research report on chatgpt

A systematic literature review of empirical research on ChatGPT in education

• Open access
• Published: 26 May 2024
• Volume 3 , article number  60 , ( 2024 )

Cite this article

You have full access to this open access article

• Yazid Albadarin   ORCID: orcid.org/0009-0005-8068-8902 1 ,
• Mohammed Saqr 1 ,
• Nicolas Pope 1 &
• Markku Tukiainen 1  

166 Accesses

Explore all metrics

Over the last four decades, studies have investigated the incorporation of Artificial Intelligence (AI) into education. A recent prominent AI-powered technology that has impacted the education sector is ChatGPT. This article provides a systematic review of 14 empirical studies incorporating ChatGPT into various educational settings, published in 2022 and before the 10th of April 2023—the date of conducting the search process. It carefully followed the essential steps outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines, as well as Okoli’s (Okoli in Commun Assoc Inf Syst, 2015) steps for conducting a rigorous and transparent systematic review. In this review, we aimed to explore how students and teachers have utilized ChatGPT in various educational settings, as well as the primary findings of those studies. By employing Creswell’s (Creswell in Educational research: planning, conducting, and evaluating quantitative and qualitative research [Ebook], Pearson Education, London, 2015) coding techniques for data extraction and interpretation, we sought to gain insight into their initial attempts at ChatGPT incorporation into education. This approach also enabled us to extract insights and considerations that can facilitate its effective and responsible use in future educational contexts. The results of this review show that learners have utilized ChatGPT as a virtual intelligent assistant, where it offered instant feedback, on-demand answers, and explanations of complex topics. Additionally, learners have used it to enhance their writing and language skills by generating ideas, composing essays, summarizing, translating, paraphrasing texts, or checking grammar. Moreover, learners turned to it as an aiding tool to facilitate their directed and personalized learning by assisting in understanding concepts and homework, providing structured learning plans, and clarifying assignments and tasks. However, the results of specific studies (n = 3, 21.4%) show that overuse of ChatGPT may negatively impact innovative capacities and collaborative learning competencies among learners. Educators, on the other hand, have utilized ChatGPT to create lesson plans, generate quizzes, and provide additional resources, which helped them enhance their productivity and efficiency and promote different teaching methodologies. Despite these benefits, the majority of the reviewed studies recommend the importance of conducting structured training, support, and clear guidelines for both learners and educators to mitigate the drawbacks. This includes developing critical evaluation skills to assess the accuracy and relevance of information provided by ChatGPT, as well as strategies for integrating human interaction and collaboration into learning activities that involve AI tools. Furthermore, they also recommend ongoing research and proactive dialogue with policymakers, stakeholders, and educational practitioners to refine and enhance the use of AI in learning environments. This review could serve as an insightful resource for practitioners who seek to integrate ChatGPT into education and stimulate further research in the field.

Similar content being viewed by others

Empowering learners with ChatGPT: insights from a systematic literature exploration

Incorporating AI in foreign language education: An investigation into ChatGPT’s effect on foreign language learners

Large language models in education: A focus on the complementary relationship between human teachers and ChatGPT

Avoid common mistakes on your manuscript.

1 Introduction

Educational technology, a rapidly evolving field, plays a crucial role in reshaping the landscape of teaching and learning [ 82 ]. One of the most transformative technological innovations of our era that has influenced the field of education is Artificial Intelligence (AI) [ 50 ]. Over the last four decades, AI in education (AIEd) has gained remarkable attention for its potential to make significant advancements in learning, instructional methods, and administrative tasks within educational settings [ 11 ]. In particular, a large language model (LLM), a type of AI algorithm that applies artificial neural networks (ANNs) and uses massively large data sets to understand, summarize, generate, and predict new content that is almost difficult to differentiate from human creations [ 79 ], has opened up novel possibilities for enhancing various aspects of education, from content creation to personalized instruction [ 35 ]. Chatbots that leverage the capabilities of LLMs to understand and generate human-like responses have also presented the capacity to enhance student learning and educational outcomes by engaging students, offering timely support, and fostering interactive learning experiences [ 46 ].

The ongoing and remarkable technological advancements in chatbots have made their use more convenient, increasingly natural and effortless, and have expanded their potential for deployment across various domains [ 70 ]. One prominent example of chatbot applications is the Chat Generative Pre-Trained Transformer, known as ChatGPT, which was introduced by OpenAI, a leading AI research lab, on November 30th, 2022. ChatGPT employs a variety of deep learning techniques to generate human-like text, with a particular focus on recurrent neural networks (RNNs). Long short-term memory (LSTM) allows it to grasp the context of the text being processed and retain information from previous inputs. Also, the transformer architecture, a neural network architecture based on the self-attention mechanism, allows it to analyze specific parts of the input, thereby enabling it to produce more natural-sounding and coherent output. Additionally, the unsupervised generative pre-training and the fine-tuning methods allow ChatGPT to generate more relevant and accurate text for specific tasks [ 31 , 62 ]. Furthermore, reinforcement learning from human feedback (RLHF), a machine learning approach that combines reinforcement learning techniques with human-provided feedback, has helped improve ChatGPT’s model by accelerating the learning process and making it significantly more efficient.

This cutting-edge natural language processing (NLP) tool is widely recognized as one of today's most advanced LLMs-based chatbots [ 70 ], allowing users to ask questions and receive detailed, coherent, systematic, personalized, convincing, and informative human-like responses [ 55 ], even within complex and ambiguous contexts [ 63 , 77 ]. ChatGPT is considered the fastest-growing technology in history: in just three months following its public launch, it amassed an estimated 120 million monthly active users [ 16 ] with an estimated 13 million daily queries [ 49 ], surpassing all other applications [ 64 ]. This remarkable growth can be attributed to the unique features and user-friendly interface that ChatGPT offers. Its intuitive design allows users to interact seamlessly with the technology, making it accessible to a diverse range of individuals, regardless of their technical expertise [ 78 ]. Additionally, its exceptional performance results from a combination of advanced algorithms, continuous enhancements, and extensive training on a diverse dataset that includes various text sources such as books, articles, websites, and online forums [ 63 ], have contributed to a more engaging and satisfying user experience [ 62 ]. These factors collectively explain its remarkable global growth and set it apart from predecessors like Bard, Bing Chat, ERNIE, and others.

In this context, several studies have explored the technological advancements of chatbots. One noteworthy recent research effort, conducted by Schöbel et al. [ 70 ], stands out for its comprehensive analysis of more than 5,000 studies on communication agents. This study offered a comprehensive overview of the historical progression and future prospects of communication agents, including ChatGPT. Moreover, other studies have focused on making comparisons, particularly between ChatGPT and alternative chatbots like Bard, Bing Chat, ERNIE, LaMDA, BlenderBot, and various others. For example, O’Leary [ 53 ] compared two chatbots, LaMDA and BlenderBot, with ChatGPT and revealed that ChatGPT outperformed both. This superiority arises from ChatGPT’s capacity to handle a wider range of questions and generate slightly varied perspectives within specific contexts. Similarly, ChatGPT exhibited an impressive ability to formulate interpretable responses that were easily understood when compared with Google's feature snippet [ 34 ]. Additionally, ChatGPT was compared to other LLMs-based chatbots, including Bard and BERT, as well as ERNIE. The findings indicated that ChatGPT exhibited strong performance in the given tasks, often outperforming the other models [ 59 ].

Furthermore, in the education context, a comprehensive study systematically compared a range of the most promising chatbots, including Bard, Bing Chat, ChatGPT, and Ernie across a multidisciplinary test that required higher-order thinking. The study revealed that ChatGPT achieved the highest score, surpassing Bing Chat and Bard [ 64 ]. Similarly, a comparative analysis was conducted to compare ChatGPT with Bard in answering a set of 30 mathematical questions and logic problems, grouped into two question sets. Set (A) is unavailable online, while Set (B) is available online. The results revealed ChatGPT's superiority in Set (A) over Bard. Nevertheless, Bard's advantage emerged in Set (B) due to its capacity to access the internet directly and retrieve answers, a capability that ChatGPT does not possess [ 57 ]. However, through these varied assessments, ChatGPT consistently highlights its exceptional prowess compared to various alternatives in the ever-evolving chatbot technology.

The widespread adoption of chatbots, especially ChatGPT, by millions of students and educators, has sparked extensive discussions regarding its incorporation into the education sector [ 64 ]. Accordingly, many scholars have contributed to the discourse, expressing both optimism and pessimism regarding the incorporation of ChatGPT into education. For example, ChatGPT has been highlighted for its capabilities in enriching the learning and teaching experience through its ability to support different learning approaches, including adaptive learning, personalized learning, and self-directed learning [ 58 , 60 , 91 ]), deliver summative and formative feedback to students and provide real-time responses to questions, increase the accessibility of information [ 22 , 40 , 43 ], foster students’ performance, engagement and motivation [ 14 , 44 , 58 ], and enhance teaching practices [ 17 , 18 , 64 , 74 ].

On the other hand, concerns have been also raised regarding its potential negative effects on learning and teaching. These include the dissemination of false information and references [ 12 , 23 , 61 , 85 ], biased reinforcement [ 47 , 50 ], compromised academic integrity [ 18 , 40 , 66 , 74 ], and the potential decline in students' skills [ 43 , 61 , 64 , 74 ]. As a result, ChatGPT has been banned in multiple countries, including Russia, China, Venezuela, Belarus, and Iran, as well as in various educational institutions in India, Italy, Western Australia, France, and the United States [ 52 , 90 ].

Clearly, the advent of chatbots, especially ChatGPT, has provoked significant controversy due to their potential impact on learning and teaching. This indicates the necessity for further exploration to gain a deeper understanding of this technology and carefully evaluate its potential benefits, limitations, challenges, and threats to education [ 79 ]. Therefore, conducting a systematic literature review will provide valuable insights into the potential prospects and obstacles linked to its incorporation into education. This systematic literature review will primarily focus on ChatGPT, driven by the aforementioned key factors outlined above.

However, the existing literature lacks a systematic literature review of empirical studies. Thus, this systematic literature review aims to address this gap by synthesizing the existing empirical studies conducted on chatbots, particularly ChatGPT, in the field of education, highlighting how ChatGPT has been utilized in educational settings, and identifying any existing gaps. This review may be particularly useful for researchers in the field and educators who are contemplating the integration of ChatGPT or any chatbot into education. The following research questions will guide this study:

What are students' and teachers' initial attempts at utilizing ChatGPT in education?

What are the main findings derived from empirical studies that have incorporated ChatGPT into learning and teaching?

2 Methodology

To conduct this study, the authors followed the essential steps of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) and Okoli’s [ 54 ] steps for conducting a systematic review. These included identifying the study’s purpose, drafting a protocol, applying a practical screening process, searching the literature, extracting relevant data, evaluating the quality of the included studies, synthesizing the studies, and ultimately writing the review. The subsequent section provides an extensive explanation of how these steps were carried out in this study.

2.1 Identify the purpose

Given the widespread adoption of ChatGPT by students and teachers for various educational purposes, often without a thorough understanding of responsible and effective use or a clear recognition of its potential impact on learning and teaching, the authors recognized the need for further exploration of ChatGPT's impact on education in this early stage. Therefore, they have chosen to conduct a systematic literature review of existing empirical studies that incorporate ChatGPT into educational settings. Despite the limited number of empirical studies due to the novelty of the topic, their goal is to gain a deeper understanding of this technology and proactively evaluate its potential benefits, limitations, challenges, and threats to education. This effort could help to understand initial reactions and attempts at incorporating ChatGPT into education and bring out insights and considerations that can inform the future development of education.

2.2 Draft the protocol

The next step is formulating the protocol. This protocol serves to outline the study process in a rigorous and transparent manner, mitigating researcher bias in study selection and data extraction [ 88 ]. The protocol will include the following steps: generating the research question, predefining a literature search strategy, identifying search locations, establishing selection criteria, assessing the studies, developing a data extraction strategy, and creating a timeline.

2.3 Apply practical screen

The screening step aims to accurately filter the articles resulting from the searching step and select the empirical studies that have incorporated ChatGPT into educational contexts, which will guide us in answering the research questions and achieving the objectives of this study. To ensure the rigorous execution of this step, our inclusion and exclusion criteria were determined based on the authors' experience and informed by previous successful systematic reviews [ 21 ]. Table 1 summarizes the inclusion and exclusion criteria for study selection.

2.4 Literature search

We conducted a thorough literature search to identify articles that explored, examined, and addressed the use of ChatGPT in Educational contexts. We utilized two research databases: Dimensions.ai, which provides access to a large number of research publications, and lens.org, which offers access to over 300 million articles, patents, and other research outputs from diverse sources. Additionally, we included three databases, Scopus, Web of Knowledge, and ERIC, which contain relevant research on the topic that addresses our research questions. To browse and identify relevant articles, we used the following search formula: ("ChatGPT" AND "Education"), which included the Boolean operator "AND" to get more specific results. The subject area in the Scopus and ERIC databases were narrowed to "ChatGPT" and "Education" keywords, and in the WoS database was limited to the "Education" category. The search was conducted between the 3rd and 10th of April 2023, which resulted in 276 articles from all selected databases (111 articles from Dimensions.ai, 65 from Scopus, 28 from Web of Science, 14 from ERIC, and 58 from Lens.org). These articles were imported into the Rayyan web-based system for analysis. The duplicates were identified automatically by the system. Subsequently, the first author manually reviewed the duplicated articles ensured that they had the same content, and then removed them, leaving us with 135 unique articles. Afterward, the titles, abstracts, and keywords of the first 40 manuscripts were scanned and reviewed by the first author and were discussed with the second and third authors to resolve any disagreements. Subsequently, the first author proceeded with the filtering process for all articles and carefully applied the inclusion and exclusion criteria as presented in Table  1 . Articles that met any one of the exclusion criteria were eliminated, resulting in 26 articles. Afterward, the authors met to carefully scan and discuss them. The authors agreed to eliminate any empirical studies solely focused on checking ChatGPT capabilities, as these studies do not guide us in addressing the research questions and achieving the study's objectives. This resulted in 14 articles eligible for analysis.

2.5 Quality appraisal

The examination and evaluation of the quality of the extracted articles is a vital step [ 9 ]. Therefore, the extracted articles were carefully evaluated for quality using Fink’s [ 24 ] standards, which emphasize the necessity for detailed descriptions of methodology, results, conclusions, strengths, and limitations. The process began with a thorough assessment of each study's design, data collection, and analysis methods to ensure their appropriateness and comprehensive execution. The clarity, consistency, and logical progression from data to results and conclusions were also critically examined. Potential biases and recognized limitations within the studies were also scrutinized. Ultimately, two articles were excluded for failing to meet Fink’s criteria, particularly in providing sufficient detail on methodology, results, conclusions, strengths, or limitations. The review process is illustrated in Fig.  1 .

The study selection process

2.6 Data extraction

The next step is data extraction, the process of capturing the key information and categories from the included studies. To improve efficiency, reduce variation among authors, and minimize errors in data analysis, the coding categories were constructed using Creswell's [ 15 ] coding techniques for data extraction and interpretation. The coding process involves three sequential steps. The initial stage encompasses open coding , where the researcher examines the data, generates codes to describe and categorize it, and gains a deeper understanding without preconceived ideas. Following open coding is axial coding , where the interrelationships between codes from open coding are analyzed to establish more comprehensive categories or themes. The process concludes with selective coding , refining and integrating categories or themes to identify core concepts emerging from the data. The first coder performed the coding process, then engaged in discussions with the second and third authors to finalize the coding categories for the first five articles. The first coder then proceeded to code all studies and engaged again in discussions with the other authors to ensure the finalization of the coding process. After a comprehensive analysis and capturing of the key information from the included studies, the data extraction and interpretation process yielded several themes. These themes have been categorized and are presented in Table  2 . It is important to note that open coding results were removed from Table  2 for aesthetic reasons, as it included many generic aspects, such as words, short phrases, or sentences mentioned in the studies.

2.7 Synthesize studies

In this stage, we will gather, discuss, and analyze the key findings that emerged from the selected studies. The synthesis stage is considered a transition from an author-centric to a concept-centric focus, enabling us to map all the provided information to achieve the most effective evaluation of the data [ 87 ]. Initially, the authors extracted data that included general information about the selected studies, including the author(s)' names, study titles, years of publication, educational levels, research methodologies, sample sizes, participants, main aims or objectives, raw data sources, and analysis methods. Following that, all key information and significant results from the selected studies were compiled using Creswell’s [ 15 ] coding techniques for data extraction and interpretation to identify core concepts and themes emerging from the data, focusing on those that directly contributed to our research questions and objectives, such as the initial utilization of ChatGPT in learning and teaching, learners' and educators' familiarity with ChatGPT, and the main findings of each study. Finally, the data related to each selected study were extracted into an Excel spreadsheet for data processing. The Excel spreadsheet was reviewed by the authors, including a series of discussions to ensure the finalization of this process and prepare it for further analysis. Afterward, the final result being analyzed and presented in various types of charts and graphs. Table 4 presents the extracted data from the selected studies, with each study labeled with a capital 'S' followed by a number.

This section consists of two main parts. The first part provides a descriptive analysis of the data compiled from the reviewed studies. The second part presents the answers to the research questions and the main findings of these studies.

3.1 Part 1: descriptive analysis

This section will provide a descriptive analysis of the reviewed studies, including educational levels and fields, participants distribution, country contribution, research methodologies, study sample size, study population, publication year, list of journals, familiarity with ChatGPT, source of data, and the main aims and objectives of the studies. Table 4 presents a comprehensive overview of the extracted data from the selected studies.

3.1.1 The number of the reviewed studies and publication years

The total number of the reviewed studies was 14. All studies were empirical studies and published in different journals focusing on Education and Technology. One study was published in 2022 [S1], while the remaining were published in 2023 [S2]-[S14]. Table 3 illustrates the year of publication, the names of the journals, and the number of reviewed studies published in each journal for the studies reviewed.

3.1.2 Educational levels and fields

The majority of the reviewed studies, 11 studies, were conducted in higher education institutions [S1]-[S10] and [S13]. Two studies did not specify the educational level of the population [S12] and [S14], while one study focused on elementary education [S11]. However, the reviewed studies covered various fields of education. Three studies focused on Arts and Humanities Education [S8], [S11], and [S14], specifically English Education. Two studies focused on Engineering Education, with one in Computer Engineering [S2] and the other in Construction Education [S3]. Two studies focused on Mathematics Education [S5] and [S12]. One study focused on Social Science Education [S13]. One study focused on Early Education [S4]. One study focused on Journalism Education [S9]. Finally, three studies did not specify the field of education [S1], [S6], and [S7]. Figure  2 represents the educational levels in the reviewed studies, while Fig.  3 represents the context of the reviewed studies.

Educational levels in the reviewed studies

Context of the reviewed studies

3.1.3 Participants distribution and countries contribution

The reviewed studies have been conducted across different geographic regions, providing a diverse representation of the studies. The majority of the studies, 10 in total, [S1]-[S3], [S5]-[S9], [S11], and [S14], primarily focused on participants from single countries such as Pakistan, the United Arab Emirates, China, Indonesia, Poland, Saudi Arabia, South Korea, Spain, Tajikistan, and the United States. In contrast, four studies, [S4], [S10], [S12], and [S13], involved participants from multiple countries, including China and the United States [S4], China, the United Kingdom, and the United States [S10], the United Arab Emirates, Oman, Saudi Arabia, and Jordan [S12], Turkey, Sweden, Canada, and Australia [ 13 ]. Figures  4 and 5 illustrate the distribution of participants, whether from single or multiple countries, and the contribution of each country in the reviewed studies, respectively.

The reviewed studies conducted in single or multiple countries

The Contribution of each country in the studies

3.1.4 Study population and sample size

Four study populations were included: university students, university teachers, university teachers and students, and elementary school teachers. Six studies involved university students [S2], [S3], [S5] and [S6]-[S8]. Three studies focused on university teachers [S1], [S4], and [S6], while one study specifically targeted elementary school teachers [S11]. Additionally, four studies included both university teachers and students [S10] and [ 12 , 13 , 14 ], and among them, study [S13] specifically included postgraduate students. In terms of the sample size of the reviewed studies, nine studies included a small sample size of less than 50 participants [S1], [S3], [S6], [S8], and [S10]-[S13]. Three studies had 50–100 participants [S2], [S9], and [S14]. Only one study had more than 100 participants [S7]. It is worth mentioning that study [S4] adopted a mixed methods approach, including 10 participants for qualitative analysis and 110 participants for quantitative analysis.

3.1.5 Participants’ familiarity with using ChatGPT

The reviewed studies recruited a diverse range of participants with varying levels of familiarity with ChatGPT. Five studies [S2], [S4], [S6], [S8], and [S12] involved participants already familiar with ChatGPT, while eight studies [S1], [S3], [S5], [S7], [S9], [S10], [S13] and [S14] included individuals with differing levels of familiarity. Notably, one study [S11] had participants who were entirely unfamiliar with ChatGPT. It is important to note that four studies [S3], [S5], [S9], and [S11] provided training or guidance to their participants before conducting their studies, while ten studies [S1], [S2], [S4], [S6]-[S8], [S10], and [S12]-[S14] did not provide training due to the participants' existing familiarity with ChatGPT.

3.1.6 Research methodology approaches and source(S) of data

The reviewed studies adopted various research methodology approaches. Seven studies adopted qualitative research methodology [S1], [S4], [S6], [S8], [S10], [S11], and [S12], while three studies adopted quantitative research methodology [S3], [S7], and [S14], and four studies employed mixed-methods, which involved a combination of both the strengths of qualitative and quantitative methods [S2], [S5], [S9], and [S13].

In terms of the source(s) of data, the reviewed studies obtained their data from various sources, such as interviews, questionnaires, and pre-and post-tests. Six studies relied on interviews as their primary source of data collection [S1], [S4], [S6], [S10], [S11], and [S12], four studies relied on questionnaires [S2], [S7], [S13], and [S14], two studies combined the use of pre-and post-tests and questionnaires for data collection [S3] and [S9], while two studies combined the use of questionnaires and interviews to obtain the data [S5] and [S8]. It is important to note that six of the reviewed studies were quasi-experimental [S3], [S5], [S8], [S9], [S12], and [S14], while the remaining ones were experimental studies [S1], [S2], [S4], [S6], [S7], [S10], [S11], and [S13]. Figures  6 and 7 illustrate the research methodologies and the source (s) of data used in the reviewed studies, respectively.

Research methodologies in the reviewed studies

Source of data in the reviewed studies

3.1.7 The aim and objectives of the studies

The reviewed studies encompassed a diverse set of aims, with several of them incorporating multiple primary objectives. Six studies [S3], [S6], [S7], [S8], [S11], and [S12] examined the integration of ChatGPT in educational contexts, and four studies [S4], [S5], [S13], and [S14] investigated the various implications of its use in education, while three studies [S2], [S9], and [S10] aimed to explore both its integration and implications in education. Additionally, seven studies explicitly explored attitudes and perceptions of students [S2] and [S3], educators [S1] and [S6], or both [S10], [S12], and [S13] regarding the utilization of ChatGPT in educational settings.

3.2 Part 2: research questions and main findings of the reviewed studies

This part will present the answers to the research questions and the main findings of the reviewed studies, classified into two main categories (learning and teaching) according to AI Education classification by [ 36 ]. Figure  8 summarizes the main findings of the reviewed studies in a visually informative diagram. Table 4 provides a detailed list of the key information extracted from the selected studies that led to generating these themes.

The main findings in the reviewed studies

4 Students' initial attempts at utilizing ChatGPT in learning and main findings from students' perspective

4.1 virtual intelligent assistant.

Nine studies demonstrated that ChatGPT has been utilized by students as an intelligent assistant to enhance and support their learning. Students employed it for various purposes, such as answering on-demand questions [S2]-[S5], [S8], [S10], and [S12], providing valuable information and learning resources [S2]-[S5], [S6], and [S8], as well as receiving immediate feedback [S2], [S4], [S9], [S10], and [S12]. In this regard, students generally were confident in the accuracy of ChatGPT's responses, considering them relevant, reliable, and detailed [S3], [S4], [S5], and [S8]. However, some students indicated the need for improvement, as they found that answers are not always accurate [S2], and that misleading information may have been provided or that it may not always align with their expectations [S6] and [S10]. It was also observed by the students that the accuracy of ChatGPT is dependent on several factors, including the quality and specificity of the user's input, the complexity of the question or topic, and the scope and relevance of its training data [S12]. Many students felt that ChatGPT's answers were not always accurate and most of them believed that it requires good background knowledge to work with.

4.2 Writing and language proficiency assistant

Six of the reviewed studies highlighted that ChatGPT has been utilized by students as a valuable assistant tool to improve their academic writing skills and language proficiency. Among these studies, three mainly focused on English education, demonstrating that students showed sufficient mastery in using ChatGPT for generating ideas, summarizing, paraphrasing texts, and completing writing essays [S8], [S11], and [S14]. Furthermore, ChatGPT helped them in writing by making students active investigators rather than passive knowledge recipients and facilitated the development of their writing skills [S11] and [S14]. Similarly, ChatGPT allowed students to generate unique ideas and perspectives, leading to deeper analysis and reflection on their journalism writing [S9]. In terms of language proficiency, ChatGPT allowed participants to translate content into their home languages, making it more accessible and relevant to their context [S4]. It also enabled them to request changes in linguistic tones or flavors [S8]. Moreover, participants used it to check grammar or as a dictionary [S11].

4.3 Valuable resource for learning approaches

Five studies demonstrated that students used ChatGPT as a valuable complementary resource for self-directed learning. It provided learning resources and guidance on diverse educational topics and created a supportive home learning environment [S2] and [S4]. Moreover, it offered step-by-step guidance to grasp concepts at their own pace and enhance their understanding [S5], streamlined task and project completion carried out independently [S7], provided comprehensive and easy-to-understand explanations on various subjects [S10], and assisted in studying geometry operations, thereby empowering them to explore geometry operations at their own pace [S12]. Three studies showed that students used ChatGPT as a valuable learning resource for personalized learning. It delivered age-appropriate conversations and tailored teaching based on a child's interests [S4], acted as a personalized learning assistant, adapted to their needs and pace, which assisted them in understanding mathematical concepts [S12], and enabled personalized learning experiences in social sciences by adapting to students' needs and learning styles [S13]. On the other hand, it is important to note that, according to one study [S5], students suggested that using ChatGPT may negatively affect collaborative learning competencies between students.

4.4 Enhancing students' competencies

Six of the reviewed studies have shown that ChatGPT is a valuable tool for improving a wide range of skills among students. Two studies have provided evidence that ChatGPT led to improvements in students' critical thinking, reasoning skills, and hazard recognition competencies through engaging them in interactive conversations or activities and providing responses related to their disciplines in journalism [S5] and construction education [S9]. Furthermore, two studies focused on mathematical education have shown the positive impact of ChatGPT on students' problem-solving abilities in unraveling problem-solving questions [S12] and enhancing the students' understanding of the problem-solving process [S5]. Lastly, one study indicated that ChatGPT effectively contributed to the enhancement of conversational social skills [S4].

4.5 Supporting students' academic success

Seven of the reviewed studies highlighted that students found ChatGPT to be beneficial for learning as it enhanced learning efficiency and improved the learning experience. It has been observed to improve students' efficiency in computer engineering studies by providing well-structured responses and good explanations [S2]. Additionally, students found it extremely useful for hazard reporting [S3], and it also enhanced their efficiency in solving mathematics problems and capabilities [S5] and [S12]. Furthermore, by finding information, generating ideas, translating texts, and providing alternative questions, ChatGPT aided students in deepening their understanding of various subjects [S6]. It contributed to an increase in students' overall productivity [S7] and improved efficiency in composing written tasks [S8]. Regarding learning experiences, ChatGPT was instrumental in assisting students in identifying hazards that they might have otherwise overlooked [S3]. It also improved students' learning experiences in solving mathematics problems and developing abilities [S5] and [S12]. Moreover, it increased students' successful completion of important tasks in their studies [S7], particularly those involving average difficulty writing tasks [S8]. Additionally, ChatGPT increased the chances of educational success by providing students with baseline knowledge on various topics [S10].

5 Teachers' initial attempts at utilizing ChatGPT in teaching and main findings from teachers' perspective

5.1 valuable resource for teaching.

The reviewed studies showed that teachers have employed ChatGPT to recommend, modify, and generate diverse, creative, organized, and engaging educational contents, teaching materials, and testing resources more rapidly [S4], [S6], [S10] and [S11]. Additionally, teachers experienced increased productivity as ChatGPT facilitated quick and accurate responses to questions, fact-checking, and information searches [S1]. It also proved valuable in constructing new knowledge [S6] and providing timely answers to students' questions in classrooms [S11]. Moreover, ChatGPT enhanced teachers' efficiency by generating new ideas for activities and preplanning activities for their students [S4] and [S6], including interactive language game partners [S11].

5.2 Improving productivity and efficiency

The reviewed studies showed that participants' productivity and work efficiency have been significantly enhanced by using ChatGPT as it enabled them to allocate more time to other tasks and reduce their overall workloads [S6], [S10], [S11], [S13], and [S14]. However, three studies [S1], [S4], and [S11], indicated a negative perception and attitude among teachers toward using ChatGPT. This negativity stemmed from a lack of necessary skills to use it effectively [S1], a limited familiarity with it [S4], and occasional inaccuracies in the content provided by it [S10].

5.3 Catalyzing new teaching methodologies

Five of the reviewed studies highlighted that educators found the necessity of redefining their teaching profession with the assistance of ChatGPT [S11], developing new effective learning strategies [S4], and adapting teaching strategies and methodologies to ensure the development of essential skills for future engineers [S5]. They also emphasized the importance of adopting new educational philosophies and approaches that can evolve with the introduction of ChatGPT into the classroom [S12]. Furthermore, updating curricula to focus on improving human-specific features, such as emotional intelligence, creativity, and philosophical perspectives [S13], was found to be essential.

5.4 Effective utilization of CHATGPT in teaching

According to the reviewed studies, effective utilization of ChatGPT in education requires providing teachers with well-structured training, support, and adequate background on how to use ChatGPT responsibly [S1], [S3], [S11], and [S12]. Establishing clear rules and regulations regarding its usage is essential to ensure it positively impacts the teaching and learning processes, including students' skills [S1], [S4], [S5], [S8], [S9], and [S11]-[S14]. Moreover, conducting further research and engaging in discussions with policymakers and stakeholders is indeed crucial for the successful integration of ChatGPT in education and to maximize the benefits for both educators and students [S1], [S6]-[S10], and [S12]-[S14].

6 Discussion

The purpose of this review is to conduct a systematic review of empirical studies that have explored the utilization of ChatGPT, one of today’s most advanced LLM-based chatbots, in education. The findings of the reviewed studies showed several ways of ChatGPT utilization in different learning and teaching practices as well as it provided insights and considerations that can facilitate its effective and responsible use in future educational contexts. The results of the reviewed studies came from diverse fields of education, which helped us avoid a biased review that is limited to a specific field. Similarly, the reviewed studies have been conducted across different geographic regions. This kind of variety in geographic representation enriched the findings of this review.

In response to RQ1 , "What are students' and teachers' initial attempts at utilizing ChatGPT in education?", the findings from this review provide comprehensive insights. Chatbots, including ChatGPT, play a crucial role in supporting student learning, enhancing their learning experiences, and facilitating diverse learning approaches [ 42 , 43 ]. This review found that this tool, ChatGPT, has been instrumental in enhancing students' learning experiences by serving as a virtual intelligent assistant, providing immediate feedback, on-demand answers, and engaging in educational conversations. Additionally, students have benefited from ChatGPT’s ability to generate ideas, compose essays, and perform tasks like summarizing, translating, paraphrasing texts, or checking grammar, thereby enhancing their writing and language competencies. Furthermore, students have turned to ChatGPT for assistance in understanding concepts and homework, providing structured learning plans, and clarifying assignments and tasks, which fosters a supportive home learning environment, allowing them to take responsibility for their own learning and cultivate the skills and approaches essential for supportive home learning environment [ 26 , 27 , 28 ]. This finding aligns with the study of Saqr et al. [ 68 , 69 ] who highlighted that, when students actively engage in their own learning process, it yields additional advantages, such as heightened motivation, enhanced achievement, and the cultivation of enthusiasm, turning them into advocates for their own learning.

Moreover, students have utilized ChatGPT for tailored teaching and step-by-step guidance on diverse educational topics, streamlining task and project completion, and generating and recommending educational content. This personalization enhances the learning environment, leading to increased academic success. This finding aligns with other recent studies [ 26 , 27 , 28 , 60 , 66 ] which revealed that ChatGPT has the potential to offer personalized learning experiences and support an effective learning process by providing students with customized feedback and explanations tailored to their needs and abilities. Ultimately, fostering students' performance, engagement, and motivation, leading to increase students' academic success [ 14 , 44 , 58 ]. This ultimate outcome is in line with the findings of Saqr et al. [ 68 , 69 ], which emphasized that learning strategies are important catalysts of students' learning, as students who utilize effective learning strategies are more likely to have better academic achievement.

Teachers, too, have capitalized on ChatGPT's capabilities to enhance productivity and efficiency, using it for creating lesson plans, generating quizzes, providing additional resources, generating and preplanning new ideas for activities, and aiding in answering students’ questions. This adoption of technology introduces new opportunities to support teaching and learning practices, enhancing teacher productivity. This finding aligns with those of Day [ 17 ], De Castro [ 18 ], and Su and Yang [ 74 ] as well as with those of Valtonen et al. [ 82 ], who revealed that emerging technological advancements have opened up novel opportunities and means to support teaching and learning practices, and enhance teachers’ productivity.

In response to RQ2 , "What are the main findings derived from empirical studies that have incorporated ChatGPT into learning and teaching?", the findings from this review provide profound insights and raise significant concerns. Starting with the insights, chatbots, including ChatGPT, have demonstrated the potential to reshape and revolutionize education, creating new, novel opportunities for enhancing the learning process and outcomes [ 83 ], facilitating different learning approaches, and offering a range of pedagogical benefits [ 19 , 43 , 72 ]. In this context, this review found that ChatGPT could open avenues for educators to adopt or develop new effective learning and teaching strategies that can evolve with the introduction of ChatGPT into the classroom. Nonetheless, there is an evident lack of research understanding regarding the potential impact of generative machine learning models within diverse educational settings [ 83 ]. This necessitates teachers to attain a high level of proficiency in incorporating chatbots, such as ChatGPT, into their classrooms to create inventive, well-structured, and captivating learning strategies. In the same vein, the review also found that teachers without the requisite skills to utilize ChatGPT realized that it did not contribute positively to their work and could potentially have adverse effects [ 37 ]. This concern could lead to inequity of access to the benefits of chatbots, including ChatGPT, as individuals who lack the necessary expertise may not be able to harness their full potential, resulting in disparities in educational outcomes and opportunities. Therefore, immediate action is needed to address these potential issues. A potential solution is offering training, support, and competency development for teachers to ensure that all of them can leverage chatbots, including ChatGPT, effectively and equitably in their educational practices [ 5 , 28 , 80 ], which could enhance accessibility and inclusivity, and potentially result in innovative outcomes [ 82 , 83 ].

Additionally, chatbots, including ChatGPT, have the potential to significantly impact students' thinking abilities, including retention, reasoning, analysis skills [ 19 , 45 ], and foster innovation and creativity capabilities [ 83 ]. This review found that ChatGPT could contribute to improving a wide range of skills among students. However, it found that frequent use of ChatGPT may result in a decrease in innovative capacities, collaborative skills and cognitive capacities, and students' motivation to attend classes, as well as could lead to reduced higher-order thinking skills among students [ 22 , 29 ]. Therefore, immediate action is needed to carefully examine the long-term impact of chatbots such as ChatGPT, on learning outcomes as well as to explore its incorporation into educational settings as a supportive tool without compromising students' cognitive development and critical thinking abilities. In the same vein, the review also found that it is challenging to draw a consistent conclusion regarding the potential of ChatGPT to aid self-directed learning approach. This finding aligns with the recent study of Baskara [ 8 ]. Therefore, further research is needed to explore the potential of ChatGPT for self-directed learning. One potential solution involves utilizing learning analytics as a novel approach to examine various aspects of students' learning and support them in their individual endeavors [ 32 ]. This approach can bridge this gap by facilitating an in-depth analysis of how learners engage with ChatGPT, identifying trends in self-directed learning behavior, and assessing its influence on their outcomes.

Turning to the significant concerns, on the other hand, a fundamental challenge with LLM-based chatbots, including ChatGPT, is the accuracy and quality of the provided information and responses, as they provide false information as truth—a phenomenon often referred to as "hallucination" [ 3 , 49 ]. In this context, this review found that the provided information was not entirely satisfactory. Consequently, the utilization of chatbots presents potential concerns, such as generating and providing inaccurate or misleading information, especially for students who utilize it to support their learning. This finding aligns with other findings [ 6 , 30 , 35 , 40 ] which revealed that incorporating chatbots such as ChatGPT, into education presents challenges related to its accuracy and reliability due to its training on a large corpus of data, which may contain inaccuracies and the way users formulate or ask ChatGPT. Therefore, immediate action is needed to address these potential issues. One possible solution is to equip students with the necessary skills and competencies, which include a background understanding of how to use it effectively and the ability to assess and evaluate the information it generates, as the accuracy and the quality of the provided information depend on the input, its complexity, the topic, and the relevance of its training data [ 28 , 49 , 86 ]. However, it's also essential to examine how learners can be educated about how these models operate, the data used in their training, and how to recognize their limitations, challenges, and issues [ 79 ].

Furthermore, chatbots present a substantial challenge concerning maintaining academic integrity [ 20 , 56 ] and copyright violations [ 83 ], which are significant concerns in education. The review found that the potential misuse of ChatGPT might foster cheating, facilitate plagiarism, and threaten academic integrity. This issue is also affirmed by the research conducted by Basic et al. [ 7 ], who presented evidence that students who utilized ChatGPT in their writing assignments had more plagiarism cases than those who did not. These findings align with the conclusions drawn by Cotton et al. [ 13 ], Hisan and Amri [ 33 ] and Sullivan et al. [ 75 ], who revealed that the integration of chatbots such as ChatGPT into education poses a significant challenge to the preservation of academic integrity. Moreover, chatbots, including ChatGPT, have increased the difficulty in identifying plagiarism [ 47 , 67 , 76 ]. The findings from previous studies [ 1 , 84 ] indicate that AI-generated text often went undetected by plagiarism software, such as Turnitin. However, Turnitin and other similar plagiarism detection tools, such as ZeroGPT, GPTZero, and Copyleaks, have since evolved, incorporating enhanced techniques to detect AI-generated text, despite the possibility of false positives, as noted in different studies that have found these tools still not yet fully ready to accurately and reliably identify AI-generated text [ 10 , 51 ], and new novel detection methods may need to be created and implemented for AI-generated text detection [ 4 ]. This potential issue could lead to another concern, which is the difficulty of accurately evaluating student performance when they utilize chatbots such as ChatGPT assistance in their assignments. Consequently, the most LLM-driven chatbots present a substantial challenge to traditional assessments [ 64 ]. The findings from previous studies indicate the importance of rethinking, improving, and redesigning innovative assessment methods in the era of chatbots [ 14 , 20 , 64 , 75 ]. These methods should prioritize the process of evaluating students' ability to apply knowledge to complex cases and demonstrate comprehension, rather than solely focusing on the final product for assessment. Therefore, immediate action is needed to address these potential issues. One possible solution would be the development of clear guidelines, regulatory policies, and pedagogical guidance. These measures would help regulate the proper and ethical utilization of chatbots, such as ChatGPT, and must be established before their introduction to students [ 35 , 38 , 39 , 41 , 89 ].

In summary, our review has delved into the utilization of ChatGPT, a prominent example of chatbots, in education, addressing the question of how ChatGPT has been utilized in education. However, there remain significant gaps, which necessitate further research to shed light on this area.

7 Conclusions

This systematic review has shed light on the varied initial attempts at incorporating ChatGPT into education by both learners and educators, while also offering insights and considerations that can facilitate its effective and responsible use in future educational contexts. From the analysis of 14 selected studies, the review revealed the dual-edged impact of ChatGPT in educational settings. On the positive side, ChatGPT significantly aided the learning process in various ways. Learners have used it as a virtual intelligent assistant, benefiting from its ability to provide immediate feedback, on-demand answers, and easy access to educational resources. Additionally, it was clear that learners have used it to enhance their writing and language skills, engaging in practices such as generating ideas, composing essays, and performing tasks like summarizing, translating, paraphrasing texts, or checking grammar. Importantly, other learners have utilized it in supporting and facilitating their directed and personalized learning on a broad range of educational topics, assisting in understanding concepts and homework, providing structured learning plans, and clarifying assignments and tasks. Educators, on the other hand, found ChatGPT beneficial for enhancing productivity and efficiency. They used it for creating lesson plans, generating quizzes, providing additional resources, and answers learners' questions, which saved time and allowed for more dynamic and engaging teaching strategies and methodologies.

However, the review also pointed out negative impacts. The results revealed that overuse of ChatGPT could decrease innovative capacities and collaborative learning among learners. Specifically, relying too much on ChatGPT for quick answers can inhibit learners' critical thinking and problem-solving skills. Learners might not engage deeply with the material or consider multiple solutions to a problem. This tendency was particularly evident in group projects, where learners preferred consulting ChatGPT individually for solutions over brainstorming and collaborating with peers, which negatively affected their teamwork abilities. On a broader level, integrating ChatGPT into education has also raised several concerns, including the potential for providing inaccurate or misleading information, issues of inequity in access, challenges related to academic integrity, and the possibility of misusing the technology.

Accordingly, this review emphasizes the urgency of developing clear rules, policies, and regulations to ensure ChatGPT's effective and responsible use in educational settings, alongside other chatbots, by both learners and educators. This requires providing well-structured training to educate them on responsible usage and understanding its limitations, along with offering sufficient background information. Moreover, it highlights the importance of rethinking, improving, and redesigning innovative teaching and assessment methods in the era of ChatGPT. Furthermore, conducting further research and engaging in discussions with policymakers and stakeholders are essential steps to maximize the benefits for both educators and learners and ensure academic integrity.

It is important to acknowledge that this review has certain limitations. Firstly, the limited inclusion of reviewed studies can be attributed to several reasons, including the novelty of the technology, as new technologies often face initial skepticism and cautious adoption; the lack of clear guidelines or best practices for leveraging this technology for educational purposes; and institutional or governmental policies affecting the utilization of this technology in educational contexts. These factors, in turn, have affected the number of studies available for review. Secondly, the utilization of the original version of ChatGPT, based on GPT-3 or GPT-3.5, implies that new studies utilizing the updated version, GPT-4 may lead to different findings. Therefore, conducting follow-up systematic reviews is essential once more empirical studies on ChatGPT are published. Additionally, long-term studies are necessary to thoroughly examine and assess the impact of ChatGPT on various educational practices.

Despite these limitations, this systematic review has highlighted the transformative potential of ChatGPT in education, revealing its diverse utilization by learners and educators alike and summarized the benefits of incorporating it into education, as well as the forefront critical concerns and challenges that must be addressed to facilitate its effective and responsible use in future educational contexts. This review could serve as an insightful resource for practitioners who seek to integrate ChatGPT into education and stimulate further research in the field.

Data availability

The data supporting our findings are available upon request.

Abbreviations

• Artificial intelligence

AI in education

Large language model

Artificial neural networks

Chat Generative Pre-Trained Transformer

Recurrent neural networks

Long short-term memory

Reinforcement learning from human feedback

Natural language processing

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

AlAfnan MA, Dishari S, Jovic M, Lomidze K. ChatGPT as an educational tool: opportunities, challenges, and recommendations for communication, business writing, and composition courses. J Artif Intell Technol. 2023. https://doi.org/10.37965/jait.2023.0184 .

Article   Google Scholar  

Ali JKM, Shamsan MAA, Hezam TA, Mohammed AAQ. Impact of ChatGPT on learning motivation. J Engl Stud Arabia Felix. 2023;2(1):41–9. https://doi.org/10.56540/jesaf.v2i1.51 .

Alkaissi H, McFarlane SI. Artificial hallucinations in ChatGPT: implications in scientific writing. Cureus. 2023. https://doi.org/10.7759/cureus.35179 .

Anderson N, Belavý DL, Perle SM, Hendricks S, Hespanhol L, Verhagen E, Memon AR. AI did not write this manuscript, or did it? Can we trick the AI text detector into generated texts? The potential future of ChatGPT and AI in sports & exercise medicine manuscript generation. BMJ Open Sport Exerc Med. 2023;9(1): e001568. https://doi.org/10.1136/bmjsem-2023-001568 .

Ausat AMA, Massang B, Efendi M, Nofirman N, Riady Y. Can chat GPT replace the role of the teacher in the classroom: a fundamental analysis. J Educ. 2023;5(4):16100–6.

Google Scholar  

Baidoo-Anu D, Ansah L. Education in the Era of generative artificial intelligence (AI): understanding the potential benefits of ChatGPT in promoting teaching and learning. Soc Sci Res Netw. 2023. https://doi.org/10.2139/ssrn.4337484 .

Basic Z, Banovac A, Kruzic I, Jerkovic I. Better by you, better than me, chatgpt3 as writing assistance in students essays. 2023. arXiv preprint arXiv:2302.04536 .‏

Baskara FR. The promises and pitfalls of using chat GPT for self-determined learning in higher education: an argumentative review. Prosiding Seminar Nasional Fakultas Tarbiyah dan Ilmu Keguruan IAIM Sinjai. 2023;2:95–101. https://doi.org/10.47435/sentikjar.v2i0.1825 .

Behera RK, Bala PK, Dhir A. The emerging role of cognitive computing in healthcare: a systematic literature review. Int J Med Inform. 2019;129:154–66. https://doi.org/10.1016/j.ijmedinf.2019.04.024 .

Chaka C. Detecting AI content in responses generated by ChatGPT, YouChat, and Chatsonic: the case of five AI content detection tools. J Appl Learn Teach. 2023. https://doi.org/10.37074/jalt.2023.6.2.12 .

Chiu TKF, Xia Q, Zhou X, Chai CS, Cheng M. Systematic literature review on opportunities, challenges, and future research recommendations of artificial intelligence in education. Comput Educ Artif Intell. 2023;4:100118. https://doi.org/10.1016/j.caeai.2022.100118 .

Choi EPH, Lee JJ, Ho M, Kwok JYY, Lok KYW. Chatting or cheating? The impacts of ChatGPT and other artificial intelligence language models on nurse education. Nurse Educ Today. 2023;125:105796. https://doi.org/10.1016/j.nedt.2023.105796 .

Cotton D, Cotton PA, Shipway JR. Chatting and cheating: ensuring academic integrity in the era of ChatGPT. Innov Educ Teach Int. 2023. https://doi.org/10.1080/14703297.2023.2190148 .

Crawford J, Cowling M, Allen K. Leadership is needed for ethical ChatGPT: Character, assessment, and learning using artificial intelligence (AI). J Univ Teach Learn Pract. 2023. https://doi.org/10.53761/1.20.3.02 .

Creswell JW. Educational research: planning, conducting, and evaluating quantitative and qualitative research [Ebook]. 4th ed. London: Pearson Education; 2015.

Curry D. ChatGPT Revenue and Usage Statistics (2023)—Business of Apps. 2023. https://www.businessofapps.com/data/chatgpt-statistics/

Day T. A preliminary investigation of fake peer-reviewed citations and references generated by ChatGPT. Prof Geogr. 2023. https://doi.org/10.1080/00330124.2023.2190373 .

De Castro CA. A Discussion about the Impact of ChatGPT in education: benefits and concerns. J Bus Theor Pract. 2023;11(2):p28. https://doi.org/10.22158/jbtp.v11n2p28 .

Deng X, Yu Z. A meta-analysis and systematic review of the effect of Chatbot technology use in sustainable education. Sustainability. 2023;15(4):2940. https://doi.org/10.3390/su15042940 .

Eke DO. ChatGPT and the rise of generative AI: threat to academic integrity? J Responsib Technol. 2023;13:100060. https://doi.org/10.1016/j.jrt.2023.100060 .

Elmoazen R, Saqr M, Tedre M, Hirsto L. A systematic literature review of empirical research on epistemic network analysis in education. IEEE Access. 2022;10:17330–48. https://doi.org/10.1109/access.2022.3149812 .

Farrokhnia M, Banihashem SK, Noroozi O, Wals AEJ. A SWOT analysis of ChatGPT: implications for educational practice and research. Innov Educ Teach Int. 2023. https://doi.org/10.1080/14703297.2023.2195846 .

Fergus S, Botha M, Ostovar M. Evaluating academic answers generated using ChatGPT. J Chem Educ. 2023;100(4):1672–5. https://doi.org/10.1021/acs.jchemed.3c00087 .

Fink A. Conducting research literature reviews: from the Internet to Paper. Incorporated: SAGE Publications; 2010.

Firaina R, Sulisworo D. Exploring the usage of ChatGPT in higher education: frequency and impact on productivity. Buletin Edukasi Indonesia (BEI). 2023;2(01):39–46. https://doi.org/10.56741/bei.v2i01.310 .

Firat, M. (2023). How chat GPT can transform autodidactic experiences and open education.  Department of Distance Education, Open Education Faculty, Anadolu Unive .‏ https://orcid.org/0000-0001-8707-5918

Firat M. What ChatGPT means for universities: perceptions of scholars and students. J Appl Learn Teach. 2023. https://doi.org/10.37074/jalt.2023.6.1.22 .

Fuchs K. Exploring the opportunities and challenges of NLP models in higher education: is Chat GPT a blessing or a curse? Front Educ. 2023. https://doi.org/10.3389/feduc.2023.1166682 .

García-Peñalvo FJ. La percepción de la inteligencia artificial en contextos educativos tras el lanzamiento de ChatGPT: disrupción o pánico. Educ Knowl Soc. 2023;24: e31279. https://doi.org/10.14201/eks.31279 .

Gilson A, Safranek CW, Huang T, Socrates V, Chi L, Taylor A, Chartash D. How does ChatGPT perform on the United States medical Licensing examination? The implications of large language models for medical education and knowledge assessment. JMIR Med Educ. 2023;9: e45312. https://doi.org/10.2196/45312 .

Hashana AJ, Brundha P, Ayoobkhan MUA, Fazila S. Deep Learning in ChatGPT—A Survey. In   2023 7th international conference on trends in electronics and informatics (ICOEI) . 2023. (pp. 1001–1005). IEEE. https://doi.org/10.1109/icoei56765.2023.10125852

Hirsto L, Saqr M, López-Pernas S, Valtonen T. (2022). A systematic narrative review of learning analytics research in K-12 and schools.  Proceedings . https://ceur-ws.org/Vol-3383/FLAIEC22_paper_9536.pdf

Hisan UK, Amri MM. ChatGPT and medical education: a double-edged sword. J Pedag Educ Sci. 2023;2(01):71–89. https://doi.org/10.13140/RG.2.2.31280.23043/1 .

Hopkins AM, Logan JM, Kichenadasse G, Sorich MJ. Artificial intelligence chatbots will revolutionize how cancer patients access information: ChatGPT represents a paradigm-shift. JNCI Cancer Spectr. 2023. https://doi.org/10.1093/jncics/pkad010 .

Househ M, AlSaad R, Alhuwail D, Ahmed A, Healy MG, Latifi S, Sheikh J. Large Language models in medical education: opportunities, challenges, and future directions. JMIR Med Educ. 2023;9: e48291. https://doi.org/10.2196/48291 .

Ilkka T. The impact of artificial intelligence on learning, teaching, and education. Minist de Educ. 2018. https://doi.org/10.2760/12297 .

Iqbal N, Ahmed H, Azhar KA. Exploring teachers’ attitudes towards using CHATGPT. Globa J Manag Adm Sci. 2022;3(4):97–111. https://doi.org/10.46568/gjmas.v3i4.163 .

Irfan M, Murray L, Ali S. Integration of Artificial intelligence in academia: a case study of critical teaching and learning in Higher education. Globa Soc Sci Rev. 2023;8(1):352–64. https://doi.org/10.31703/gssr.2023(viii-i).32 .

Jeon JH, Lee S. Large language models in education: a focus on the complementary relationship between human teachers and ChatGPT. Educ Inf Technol. 2023. https://doi.org/10.1007/s10639-023-11834-1 .

Khan RA, Jawaid M, Khan AR, Sajjad M. ChatGPT—Reshaping medical education and clinical management. Pak J Med Sci. 2023. https://doi.org/10.12669/pjms.39.2.7653 .

King MR. A conversation on artificial intelligence, Chatbots, and plagiarism in higher education. Cell Mol Bioeng. 2023;16(1):1–2. https://doi.org/10.1007/s12195-022-00754-8 .

Kooli C. Chatbots in education and research: a critical examination of ethical implications and solutions. Sustainability. 2023;15(7):5614. https://doi.org/10.3390/su15075614 .

Kuhail MA, Alturki N, Alramlawi S, Alhejori K. Interacting with educational chatbots: a systematic review. Educ Inf Technol. 2022;28(1):973–1018. https://doi.org/10.1007/s10639-022-11177-3 .

Lee H. The rise of ChatGPT: exploring its potential in medical education. Anat Sci Educ. 2023. https://doi.org/10.1002/ase.2270 .

Li L, Subbareddy R, Raghavendra CG. AI intelligence Chatbot to improve students learning in the higher education platform. J Interconnect Netw. 2022. https://doi.org/10.1142/s0219265921430325 .

Limna P. A Review of Artificial Intelligence (AI) in Education during the Digital Era. 2022. https://ssrn.com/abstract=4160798

Lo CK. What is the impact of ChatGPT on education? A rapid review of the literature. Educ Sci. 2023;13(4):410. https://doi.org/10.3390/educsci13040410 .

Luo W, He H, Liu J, Berson IR, Berson MJ, Zhou Y, Li H. Aladdin’s genie or pandora’s box For early childhood education? Experts chat on the roles, challenges, and developments of ChatGPT. Early Educ Dev. 2023. https://doi.org/10.1080/10409289.2023.2214181 .

Meyer JG, Urbanowicz RJ, Martin P, O’Connor K, Li R, Peng P, Moore JH. ChatGPT and large language models in academia: opportunities and challenges. Biodata Min. 2023. https://doi.org/10.1186/s13040-023-00339-9 .

Mhlanga D. Open AI in education, the responsible and ethical use of ChatGPT towards lifelong learning. Soc Sci Res Netw. 2023. https://doi.org/10.2139/ssrn.4354422 .

Neumann, M., Rauschenberger, M., & Schön, E. M. (2023). “We Need To Talk About ChatGPT”: The Future of AI and Higher Education.‏ https://doi.org/10.1109/seeng59157.2023.00010

Nolan B. Here are the schools and colleges that have banned the use of ChatGPT over plagiarism and misinformation fears. Business Insider . 2023. https://www.businessinsider.com

O’Leary DE. An analysis of three chatbots: BlenderBot, ChatGPT and LaMDA. Int J Intell Syst Account, Financ Manag. 2023;30(1):41–54. https://doi.org/10.1002/isaf.1531 .

Okoli C. A guide to conducting a standalone systematic literature review. Commun Assoc Inf Syst. 2015. https://doi.org/10.17705/1cais.03743 .

OpenAI. (2023). https://openai.com/blog/chatgpt

Perkins M. Academic integrity considerations of AI large language models in the post-pandemic era: ChatGPT and beyond. J Univ Teach Learn Pract. 2023. https://doi.org/10.53761/1.20.02.07 .

Plevris V, Papazafeiropoulos G, Rios AJ. Chatbots put to the test in math and logic problems: A preliminary comparison and assessment of ChatGPT-3.5, ChatGPT-4, and Google Bard. arXiv (Cornell University) . 2023. https://doi.org/10.48550/arxiv.2305.18618

Rahman MM, Watanobe Y (2023) ChatGPT for education and research: opportunities, threats, and strategies. Appl Sci 13(9):5783. https://doi.org/10.3390/app13095783

Ram B, Verma P. Artificial intelligence AI-based Chatbot study of ChatGPT, google AI bard and baidu AI. World J Adv Eng Technol Sci. 2023;8(1):258–61. https://doi.org/10.30574/wjaets.2023.8.1.0045 .

Rasul T, Nair S, Kalendra D, Robin M, de Oliveira Santini F, Ladeira WJ, Heathcote L. The role of ChatGPT in higher education: benefits, challenges, and future research directions. J Appl Learn Teach. 2023. https://doi.org/10.37074/jalt.2023.6.1.29 .

Ratnam M, Sharm B, Tomer A. ChatGPT: educational artificial intelligence. Int J Adv Trends Comput Sci Eng. 2023;12(2):84–91. https://doi.org/10.30534/ijatcse/2023/091222023 .

Ray PP. ChatGPT: a comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope. Internet Things Cyber-Phys Syst. 2023;3:121–54. https://doi.org/10.1016/j.iotcps.2023.04.003 .

Roumeliotis KI, Tselikas ND. ChatGPT and Open-AI models: a preliminary review. Future Internet. 2023;15(6):192. https://doi.org/10.3390/fi15060192 .

Rudolph J, Tan S, Tan S. War of the chatbots: Bard, Bing Chat, ChatGPT, Ernie and beyond. The new AI gold rush and its impact on higher education. J Appl Learn Teach. 2023. https://doi.org/10.37074/jalt.2023.6.1.23 .

Ruiz LMS, Moll-López S, Nuñez-Pérez A, Moraño J, Vega-Fleitas E. ChatGPT challenges blended learning methodologies in engineering education: a case study in mathematics. Appl Sci. 2023;13(10):6039. https://doi.org/10.3390/app13106039 .

Sallam M, Salim NA, Barakat M, Al-Tammemi AB. ChatGPT applications in medical, dental, pharmacy, and public health education: a descriptive study highlighting the advantages and limitations. Narra J. 2023;3(1): e103. https://doi.org/10.52225/narra.v3i1.103 .

Salvagno M, Taccone FS, Gerli AG. Can artificial intelligence help for scientific writing? Crit Care. 2023. https://doi.org/10.1186/s13054-023-04380-2 .

Saqr M, López-Pernas S, Helske S, Hrastinski S. The longitudinal association between engagement and achievement varies by time, students’ profiles, and achievement state: a full program study. Comput Educ. 2023;199:104787. https://doi.org/10.1016/j.compedu.2023.104787 .

Saqr M, Matcha W, Uzir N, Jovanović J, Gašević D, López-Pernas S. Transferring effective learning strategies across learning contexts matters: a study in problem-based learning. Australas J Educ Technol. 2023;39(3):9.

Schöbel S, Schmitt A, Benner D, Saqr M, Janson A, Leimeister JM. Charting the evolution and future of conversational agents: a research agenda along five waves and new frontiers. Inf Syst Front. 2023. https://doi.org/10.1007/s10796-023-10375-9 .

Shoufan A. Exploring students’ perceptions of CHATGPT: thematic analysis and follow-up survey. IEEE Access. 2023. https://doi.org/10.1109/access.2023.3268224 .

Sonderegger S, Seufert S. Chatbot-mediated learning: conceptual framework for the design of Chatbot use cases in education. Gallen: Institute for Educational Management and Technologies, University of St; 2022. https://doi.org/10.5220/0010999200003182 .

Book   Google Scholar  

Strzelecki A. To use or not to use ChatGPT in higher education? A study of students’ acceptance and use of technology. Interact Learn Environ. 2023. https://doi.org/10.1080/10494820.2023.2209881 .

Su J, Yang W. Unlocking the power of ChatGPT: a framework for applying generative AI in education. ECNU Rev Educ. 2023. https://doi.org/10.1177/20965311231168423 .

Sullivan M, Kelly A, McLaughlan P. ChatGPT in higher education: Considerations for academic integrity and student learning. J ApplLearn Teach. 2023;6(1):1–10. https://doi.org/10.37074/jalt.2023.6.1.17 .

Szabo A. ChatGPT is a breakthrough in science and education but fails a test in sports and exercise psychology. Balt J Sport Health Sci. 2023;1(128):25–40. https://doi.org/10.33607/bjshs.v127i4.1233 .

Taecharungroj V. “What can ChatGPT do?” analyzing early reactions to the innovative AI chatbot on Twitter. Big Data Cognit Comput. 2023;7(1):35. https://doi.org/10.3390/bdcc7010035 .

Tam S, Said RB. User preferences for ChatGPT-powered conversational interfaces versus traditional methods. Biomed Eng Soc. 2023. https://doi.org/10.58496/mjcsc/2023/004 .

Tedre M, Kahila J, Vartiainen H. (2023). Exploration on how co-designing with AI facilitates critical evaluation of ethics of AI in craft education. In: Langran E, Christensen P, Sanson J (Eds).  Proceedings of Society for Information Technology and Teacher Education International Conference . 2023. pp. 2289–2296.

Tlili A, Shehata B, Adarkwah MA, Bozkurt A, Hickey DT, Huang R, Agyemang B. What if the devil is my guardian angel: ChatGPT as a case study of using chatbots in education. Smart Learn Environ. 2023. https://doi.org/10.1186/s40561-023-00237-x .

Uddin SMJ, Albert A, Ovid A, Alsharef A. Leveraging CHATGPT to aid construction hazard recognition and support safety education and training. Sustainability. 2023;15(9):7121. https://doi.org/10.3390/su15097121 .

Valtonen T, López-Pernas S, Saqr M, Vartiainen H, Sointu E, Tedre M. The nature and building blocks of educational technology research. Comput Hum Behav. 2022;128:107123. https://doi.org/10.1016/j.chb.2021.107123 .

Vartiainen H, Tedre M. Using artificial intelligence in craft education: crafting with text-to-image generative models. Digit Creat. 2023;34(1):1–21. https://doi.org/10.1080/14626268.2023.2174557 .

Ventayen RJM. OpenAI ChatGPT generated results: similarity index of artificial intelligence-based contents. Soc Sci Res Netw. 2023. https://doi.org/10.2139/ssrn.4332664 .

Wagner MW, Ertl-Wagner BB. Accuracy of information and references using ChatGPT-3 for retrieval of clinical radiological information. Can Assoc Radiol J. 2023. https://doi.org/10.1177/08465371231171125 .

Wardat Y, Tashtoush MA, AlAli R, Jarrah AM. ChatGPT: a revolutionary tool for teaching and learning mathematics. Eurasia J Math, Sci Technol Educ. 2023;19(7):em2286. https://doi.org/10.29333/ejmste/13272 .

Webster J, Watson RT. Analyzing the past to prepare for the future: writing a literature review. Manag Inf Syst Quart. 2002;26(2):3.

Xiao Y, Watson ME. Guidance on conducting a systematic literature review. J Plan Educ Res. 2017;39(1):93–112. https://doi.org/10.1177/0739456x17723971 .

Yan D. Impact of ChatGPT on learners in a L2 writing practicum: an exploratory investigation. Educ Inf Technol. 2023. https://doi.org/10.1007/s10639-023-11742-4 .

Yu H. Reflection on whether Chat GPT should be banned by academia from the perspective of education and teaching. Front Psychol. 2023;14:1181712. https://doi.org/10.3389/fpsyg.2023.1181712 .

Zhu C, Sun M, Luo J, Li T, Wang M. How to harness the potential of ChatGPT in education? Knowl Manag ELearn. 2023;15(2):133–52. https://doi.org/10.34105/j.kmel.2023.15.008 .

Download references

The paper is co-funded by the Academy of Finland (Suomen Akatemia) Research Council for Natural Sciences and Engineering for the project Towards precision education: Idiographic learning analytics (TOPEILA), Decision Number 350560.

Author information

Authors and affiliations.

School of Computing, University of Eastern Finland, 80100, Joensuu, Finland

Yazid Albadarin, Mohammed Saqr, Nicolas Pope & Markku Tukiainen

You can also search for this author in PubMed   Google Scholar

Contributions

YA contributed to the literature search, data analysis, discussion, and conclusion. Additionally, YA contributed to the manuscript’s writing, editing, and finalization. MS contributed to the study’s design, conceptualization, acquisition of funding, project administration, allocation of resources, supervision, validation, literature search, and analysis of results. Furthermore, MS contributed to the manuscript's writing, revising, and approving it in its finalized state. NP contributed to the results, and discussions, and provided supervision. NP also contributed to the writing process, revisions, and the final approval of the manuscript in its finalized state. MT contributed to the study's conceptualization, resource management, supervision, writing, revising the manuscript, and approving it.

Corresponding author

Correspondence to Yazid Albadarin .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

See Table  4

The process of synthesizing the data presented in Table  4 involved identifying the relevant studies through a search process of databases (ERIC, Scopus, Web of Knowledge, Dimensions.ai, and lens.org) using specific keywords "ChatGPT" and "education". Following this, inclusion/exclusion criteria were applied, and data extraction was performed using Creswell's [ 15 ] coding techniques to capture key information and identify common themes across the included studies.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Albadarin, Y., Saqr, M., Pope, N. et al. A systematic literature review of empirical research on ChatGPT in education. Discov Educ 3 , 60 (2024). https://doi.org/10.1007/s44217-024-00138-2

Download citation

Received : 22 October 2023

Accepted : 10 May 2024

Published : 26 May 2024

DOI : https://doi.org/10.1007/s44217-024-00138-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Large language models
• Educational technology
• Systematic review

Advertisement

• Find a journal
• Publish with us
• Track your research

Numbers, Facts and Trends Shaping Your World

Read our research on:

Full Topic List

Regions & Countries

• Publications
• Our Methods
• Short Reads
• Tools & Resources

Read Our Research On:

A majority of Americans have heard of ChatGPT, but few have tried it themselves

For the latest survey data on ChatGPT, see “Americans’ use of ChatGPT is ticking up, but few trust its election information .”

About six-in-ten U.S. adults (58%) are familiar with ChatGPT, though relatively few have tried it themselves, according to a Pew Research Center survey conducted in March. Among those who have tried ChatGPT, a majority report it has been at least somewhat useful.

ChatGPT is an open-access online chatbot that allows users to ask questions and request content. The versatility and human-like quality of its responses have captured the attention of the media , the tech industry and some members of the public . ChatGPT surpassed 100 million monthly users within two months of its public launch in late November 2022 , setting a world record as the fastest-growing web application. Due to these factors, the Center chose to ask Americans about ChatGPT specifically rather than chatbots or large language models (LLMs) more broadly.

Pew Research Center has a history of exploring Americans’ perspectives on emerging technologies and uses of artificial intelligence. The current study sought to understand Americans’ familiarity and experiences with ChatGPT. This survey was conducted among 10,701 U.S. adults from March 13 to 19, 2023. Everyone who took part in the survey is a member of the Center’s American Trends Panel (ATP), an online survey panel that is recruited through national, random sampling of residential addresses. This way nearly all U.S. adults have a chance of selection. The survey is weighted to be representative of the U.S. adult population by gender, race and ethnicity, partisan affiliation, education and other categories. Read more about the  ATP’s methodology .

Here are the questions used for this analysis , along with responses, and  its methodology .

Overall, 18% of U.S. adults have heard a lot about ChatGPT, while 39% have heard a little and 42% have heard nothing at all. But there are considerable demographic differences in awareness of this chatbot.

For example, roughly eight-in-ten adults with a postgraduate degree have heard a lot (32%) or a little (47%) about this artificial intelligence program, while 71% of those with a bachelor’s degree say the same. A smaller share of those who have some college education (59%) say they’ve heard of it. By comparison, 41% of those with a high school education or less are familiar with ChatGPT. A similar pattern emerges with household income, as a higher share of people from more affluent households are aware of this program.

In addition, Asian adults are particularly likely to have heard of ChatGPT: 78% say they have heard at least a little about it, compared with about six-in-ten White adults and roughly half of Hispanic or Black adults. Asian adults are also more than twice as likely as adults of other races to say they have heard a lot about this program.

Men are more likely than women to have heard at least a little about ChatGPT, as are adults under 30 when compared with those 30 and older.

Just 14% of U.S. adults have tried ChatGPT

How ChatGPT should be used has been hotly contested. Some people see it as a useful tool for educational and work purposes , while others feel it should only be used for entertainment .

However, few U.S. adults have themselves used ChatGPT for any purpose. Just 14% of all U.S. adults say they have used it for entertainment, to learn something new, or for their work. This lack of uptake is in line with a Pew Research Center survey from 2021 that found that Americans were more likely to express concerns than excitement about increased use of artificial intelligence in daily life.

Among the subset of Americans who have heard of ChatGPT, 19% say they have used it for entertainment and 14% have used it to learn something new. About one-in-ten adults who have heard of ChatGPT and are currently working for pay have used it at work.

White adults who have heard of ChatGPT are consistently less likely than their Asian, Hispanic or Black counterparts to have used the chatbot for fun, education or work.

Use of ChatGPT for these purposes is also closely related to age. For example, adults under 30 who have heard of ChatGPT are far more likely than those 65 and older to have used the chatbot for entertainment (31% vs. 4%).

Roughly four-in-ten Americans who have tried ChatGPT say it has been somewhat useful

While ChatGPT can be used in many ways and for many tasks, it has come under fire for sometimes failing to produce accurate answers , making up information , using real organizations’ names (including ours) to try to legitimize its claims , and accusing real people of crimes that they did not commit . These falsehoods can be extremely convincing because ChatGPT can produce eloquent prose and cite nonexistent sources that seem real even to the people it credits .

Americans’ opinions about ChatGPT’s utility are somewhat mixed. People who have used it were asked about their experience with this chatbot. Roughly a third say it has been extremely (15%) or very useful (20%), while 39% say it has been somewhat useful. Around a quarter of those who have tried it say it has been not very (21%) or not at all useful (6%).

Younger adults tend to find ChatGPT more useful than older adults. About four-in-ten adults under 50 who have used it (38%) say it was extremely or very useful, whereas only about a quarter of users 50 and older (24%) say the same.

Note: Here are the questions used for this analysis , along with responses, and  its methodology .

• Artificial Intelligence
• Emerging Technology
• Internet & Technology
• Online Search
• Platforms & Services
• Technology Adoption

Emily A. Vogels is a former research associate focusing on internet and technology at Pew Research Center .

A quarter of U.S. teachers say AI tools do more harm than good in K-12 education

Many americans think generative ai programs should credit the sources they rely on, americans’ use of chatgpt is ticking up, but few trust its election information, q&a: how we used large language models to identify guests on popular podcasts, striking findings from 2023, most popular.

1615 L St. NW, Suite 800 Washington, DC 20036 USA (+1) 202-419-4300 | Main (+1) 202-857-8562 | Fax (+1) 202-419-4372 |  Media Inquiries

Research Topics

• Email Newsletters

ABOUT PEW RESEARCH CENTER  Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes and trends shaping the world. It conducts public opinion polling, demographic research, media content analysis and other empirical social science research. Pew Research Center does not take policy positions. It is a subsidiary of  The Pew Charitable Trusts .

© 2024 Pew Research Center

Along with Stanford news and stories, show me:

• Student information
• Faculty/Staff information

We want to provide announcements, events, leadership messages and resources that are relevant to you. Your selection is stored in a browser cookie which you can remove at any time using “Clear all personalization” below.

Since its public launch in November 2022, ChatGPT has captured the world’s attention,  showing millions of users around the globe the extraordinary potential of artificial intelligence as it churns out human-sounding sounding answers to requests ranging from the practical to the surreal. It has drafted cover letters , composed lines of poetry , pretended to be William Shakespeare , crafted messages for dating app users to woo matches , and even written news articles, all with varying results .

Communication Professor Jeff Hancock studies issues around trust and deception and how AI-mediated communication impacts interpersonal relationships. (Image credit: Andrew Brodhead)

Emerging out of these promising applications are ethical dilemmas as well. In a world increasingly dominated by AI-powered tools that can mimic human natural language abilities, what does it mean to be truthful and authentic? Stanford communication scholar Jeff Hancock has been tackling this issue and the impact of AI on interpersonal relationships in his research.

Hancock argues that the Turing test era is over: Bots now sound so real that it has become impossible for people to distinguish between humans and machines in conversations, which poses huge risks for manipulation and deception at mass scale. How then can these tools be used for good and not harm is a question that has Hancock and others worried.

While he sees the potential of AI to help how people do their work more effectively, Hancock sees pitfalls as well. Ultimately, he says, our challenge will be to develop AI that supports human goals and to educate people how best to use these new technologies in effective and ethical ways.

For several years now, Hancock has been examining how AI-mediated communication is transforming – and potentially, undermining – interpersonal relationships.

This interview has been edited for length and clarity.

How do you see AI-mediated tools like ChatGPT fitting into people’s lives?

Much like the calculator didn’t replace the need to take math or for people to work the till or to be accountants, I think we will find ways of using AI-mediated communication as a tool. I think the more we think of it as an assistant or a tool that is incredibly powerful, the more we can envision how it will be useful. But it’s important to note that these systems are not ready to plug and play right off the shelf. They’re not there yet, and neither are we humans.

ChatGPT is being used by millions of people, many of whom don’t have any training or education about when it is ethical to use these systems or how to ensure that they are not causing harm.

Technically, systems like ChatGPT are genuinely far better than things that we’ve had in the past, but there are still a lot of issues. They do not currently provide accurate information at very high rates. The best ones produce useful information that’s accurate 50% or 70% of the time , though that will likely change with new versions like the imminent GPT4. They can also produce falsehoods or make stuff up – what we call “hallucinating”. It can take a lot of work to actually get them to produce something good. Prompts are difficult and produce really different responses.

What have you learned about AI and communication that might surprise people?

My lab has been really interested in the human questions around AI. A lot of people say, “I don’t know about these new bots – I’ve never used one.” But most people have experienced some kind of AI communication, the most common being the simple, smart replies in email messages that provide options such as “that sounds good,” “that’s great,” or “sorry, I can’t.” What we found is that even if you don’t use those AI-generated responses, they influence how you think. Those three options prime you. When you write an email back, you tend to write a shorter email. You tend to write a simpler email. And it just isn’t linguistically less complex: You have more positive affect, which means you use more positive emotion terms and fewer negative ones. That’s because that’s how smart replies are built: They’re very short, they’re very simple, and they’re overly positive. You don’t even have to be using these systems to actually be affected by them.

Are there any areas in particular where you see an opportunity for ChatGPT to help people be better at their job or do it more effectively?

I think there’s a whole new world of communication that this will usher in. It really feels like an inflection point. For example, there is a tremendous amount of potential in the initial levels of therapy and coaching. Say you are procrastinating at work, or you need help negotiating, or you are having some anxiety in your job. Meeting with an actual counselor or coach is difficult and can be expensive. These systems offer ways of getting access to that kind of help, at least at the initial stages of working on a problem. I think the best combination will be when ChatGPT can support an actual coach, who can then help more people more effectively.

“Questions around authenticity, deception, and trust are going to be incredibly important, and we need a lot more research to help us understand how AI will influence how we interact with other humans.” —Jeff Hancock Professor of Communication

ChatGPT could be useful at doing some of the prep work that coaches do as they try to understand their clients’ needs. Sometimes these are standard questions that a system like ChatGPT could be trained to ask the client and then synthesize for the coach. This could potentially free the coach up to engage more deeply with that client or to help more clients.

Of course, we still don’t know how people will react to having bots involved in these kinds of conversations. Some work we did a few years ago showed that people’s emotional, relational, and psychological outcomes from conversations with bots could be similar to those with a human. There is potential there, but care will certainly be needed in introducing these kinds of systems into communication like this.

One area you study is trust and deception. How do you see AI affecting the way people trust one another?

I think if the machine is helping you be you, then I think you can be authentic. AI systems can be optimized for a whole host of things and some of those can be interpersonal. You can say, “I want an online dating description and I want to come across as very funny, warm, and open.” I just did an exercise in class, getting the students to use ChatGPT to create an online dating profile, and was shocked when all the students said that ChatGPT’s description was an accurate representation of themselves! They also agreed that they would modify it a little, but that it was surprisingly good. That could really help a person, especially for people who experience communication anxiety or aren’t very good at expressing themselves.

But for the people who are trying to use those descriptions as signals of what a person is like, our usual process of impression formation breaks down because it wasn’t you who came across as very funny, warm, and open – it was a machine doing that for you. We will have a lot of responsibility about how we go about using these tools. I think it’s really about how we as humans choose to use it.

Can we trust these systems?

There are a lot of technical questions, like how the AI was developed. A widespread concern is that these systems are relying on biased data for training. But there are also deeper philosophical questions around consciousness and intention. What does it mean for a machine to be deceptive? It can lie about who it is and talk about a lived experience it did not have, which is deceptive. But most definitions of deception include an intent to mislead someone. If the system doesn’t have that intent, is it deceptive? Does it come back to the person that was asking the questions or getting the system to be deceptive? I don’t know. There are more questions than answers at this point.

You also study disinformation. What worries you about how AI can be used for nefarious purposes?

I also want to recognize that there are real dangers for misuse here. Renée DiResta , my colleague and collaborator in the Stanford Internet Observatory , has a report out that lays out some of the main ways that these systems can be a threat around misinformation, and possible solutions. As Renée and the report note, once bad actors are able to use language models to run influence campaigns, they will be able to generate massive amounts of content quite cheaply, and potentially develop novel tactics for large-scale persuasion. It has the potential to transform and exacerbate the problem of misinformation, and so we need to start working on solutions now.

What can be done to authenticate communication?

In the last year or so there have been a number of papers showing that these large language models and chatbots can no longer be differentiated from their human counterparts.

So how to authenticate communication is a big question. To come across as authentic, you could say, “This was written or partly written by AI.” If you do that, you’re being honest, but research shows that people could perceive you as less sincere – there is a paper that came out that found if you apologize and indicate that you used AI to help you write your apology, people view it as less sincere. In one of our paper s , we showed that if you apply for a job and say you used AI to help write it, people will perceive you as less competent. There are costs if you disclose you used AI, but if you don’t disclose it, are you coming across as inauthentic? These questions around authenticity, deception, and trust are going to be incredibly important, and we need a lot more research to help us understand how AI will influence how we interact with other humans.

Hancock is the founding director of the Stanford Social Media Lab and is the Harry and Norman Chandler Professor in Communication in the School of Humanities and Sciences . He is also a faculty affiliate at the Stanford Institute for Human-Centered Artificial Intelligence and is the co-director of the Cyber Policy Center and faculty director of the Stanford Internet Observatory.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Therap Adv Gastroenterol
• PMC10750546

Evaluating the role of ChatGPT in gastroenterology: a comprehensive systematic review of applications, benefits, and limitations

Division of Data-Driven and Digital Medicine (D3M), Icahn School of Medicine at Mount Sinai, New York, NY, USA

The Charles Bronfman Institute of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA

ARC Innovation Center, Sheba Medical Center at Tel Hashomer Affiliated with Tel Aviv Medical School, Tel Aviv University, Tel Aviv, Israel

Ali Sourosh

Girish n. nadkarni, kassem sharif.

Department of Gastroenterology, Sheba Medical Center at Tel Hashomer Affiliated with Tel Aviv Medical School, Tel Aviv University, Tel Aviv, Israel

Department of Gastroenterology, Sheba Medical Center at Tel Hashomer, Ramat Gan, 52621 Affiliated with Tel Aviv Medical School, Tel Aviv University, Tel Aviv, Israel

Background:

The integration of artificial intelligence (AI) into healthcare has opened new avenues for enhancing patient care and clinical research. In gastroenterology, the potential of AI tools, specifically large language models like ChatGPT, is being explored to understand their utility and effectiveness.

Objectives:

The primary goal of this systematic review is to assess the various applications, ascertain the benefits, and identify the limitations of utilizing ChatGPT within the realm of gastroenterology.

Through a systematic approach, this review aggregates findings from multiple studies to evaluate the impact of ChatGPT on the field.

Data sources and methods:

The review was based on a detailed literature search of the PubMed database, targeting research that delves into the use of ChatGPT for gastroenterological purposes. It incorporated six selected studies, which were meticulously evaluated for quality using the Joanna Briggs Institute critical appraisal instruments. The data were then synthesized narratively to encapsulate the roles, advantages, and drawbacks of ChatGPT in gastroenterology.

The investigation unearthed various roles of ChatGPT, including its use in patient education, diagnostic self-assessment, disease management, and the formulation of research queries. Notable benefits were its capability to provide pertinent recommendations, enhance communication between patients and physicians, and prompt valuable research inquiries. Nonetheless, it encountered obstacles in decoding intricate medical questions, yielded inconsistent responses at times, and exhibited limitations in generating novel content. The review also considered ethical implications.

Conclusion:

ChatGPT has demonstrated significant potential in the field of gastroenterology, especially in facilitating patient–physician interactions and managing diseases. Despite these advancements, the review underscores the necessity for ongoing refinement, customization, and ethical regulation of AI tools. These findings serve to enrich the dialog concerning AI’s role in healthcare, with a specific focus on ChatGPT’s application in gastroenterology.

Plain language summary

Checking how ChatGPT works in gastroenterology: a detailed look at its uses, advantages, and challenges

Goal We looked at how ChatGPT, a computer program, is used in the study Gastroenterology. We wanted to understand what’s good about it, what’s challenging, and how it can help doctors and patients. How We Did It We searched for articles about ChatGPT in Gastroenterology on PubMed. We found six suitable articles and checked their quality using the Joanna Briggs Institute (JBI) critical appraisal tools. Then, we put all the information together to get a clear picture. What We Found Doctors and researchers use ChatGPT in many ways. Some use it to teach patients about their health, while others use it to help patients check their symptoms or manage their conditions. It can even help come up with research questions. The good things about ChatGPT are that it gives helpful advice, makes talking between doctors and patients easier, and helps come up with research topics. But, sometimes it doesn’t understand hard medical questions, gives different answers for the same question, or lacks new ideas. There are also concerns about using it the right way. What This Means ChatGPT can be a helpful tool in Gastroenterology, especially when talking with patients and managing their health. But, there are challenges that need to be fixed. Our review helps people understand how ChatGPT can be used in health care, especially in the field of Gastroenterology.

Introduction

Artificial intelligence (AI) and large language models (LLMs), such as ChatGPT developed by OpenAI, have been increasingly recognized for their potential to revolutionize various sectors, including healthcare. 1 , 2 In medicine, and more specifically in gastroenterology, these models have shown promise as supportive tools for clinicians, enhancing patient care and improving healthcare delivery. 3 However, while the potential benefits are substantial, the application of AI in healthcare is not without its challenges and limitations. 4 , 5

ChatGPT, a conversational AI system based on the generative pre-trained transformer (GPT) architecture, has demonstrated impressive capabilities in various gastroenterological applications. These include answering common patient questions, 6 taking part in self-assessment tests, 7 and even identifying research priorities. 8 Despite these promising applications, the performance of ChatGPT in the medical domain has been inconsistent, with concerns raised about its accuracy and efficacy. 9

A recent review article 10 noted that GPT-4 could be beneficial for patient–physician communication, patient education, and continuous patient care, potentially mitigating factors related to physicians’ burnout. However, the authors highlighted key limitations and ethical considerations of this AI technology, including patient confidentiality and data security, algorithmic bias, inconsistent and inaccurate responses, plagiarism concerns, compliance with data privacy regulations, and the irreplaceable role of human judgment.

This review aims to provide an evaluation of the role of ChatGPT in gastroenterology, drawing from existing literature. By analyzing studies on the application of ChatGPT in patient communication, medical education, disease management, and research prioritization, we aim to provide a perspective on the potential and challenges of this tool in gastroenterology.

Study selection

For this systematic review, we included studies that examined the application of ChatGPT in gastroenterology. We excluded studies that focused on other AI models or other areas of healthcare.

Search strategy

We conducted a comprehensive literature search using the PubMed database. The search strategy incorporated a combination of Medical Subject Headings (MeSH) terms and keywords related to ‘ChatGPT’, and ‘Gastroenterology’. The search was limited to articles published in English. Reference lists of included studies and relevant reviews were also manually searched to identify any additional studies.

Data extraction

Two independent reviewers extracted data from the included studies using a standardized data extraction form. Discrepancies were resolved through discussion or consultation with a third reviewer. The extracted information included: study design, sample size, application of ChatGPT (e.g. patient education, self-assessment, continuous care), main findings, and limitations.

Quality assessment

The quality of the included studies was assessed using the Joanna Briggs Institute (JBI) critical appraisal tools, appropriate to each study design. These tools assess the methodological quality of a study and the extent to which a study has addressed the possibility of bias in its design, conduct, and analysis. Studies were categorized as high, moderate, or low quality based on their scores. According to the JBI guidelines, it is recommended that critical appraisal be undertaken by at least two independent reviewers to minimize potential bias. In our study, the quality assessment using the modified JBI critical appraisal tools was conducted with the agreement of three authors (AL, EK, and KS) to ensure the objectivity and robustness of the evaluation.

Data synthesis

We conducted a narrative synthesis of the findings from the included studies. Due to the anticipated heterogeneity in study designs and outcomes, a meta-analysis was not planned. Instead, we focused on summarizing the applications, benefits, and limitations of ChatGPT in gastroenterology as reported in the studies, and on identifying areas for future research.

The systematic review included six studies that evaluated the application, benefits, and limitations of ChatGPT in the field of gastroenterology. The studies were diverse in their objectives and methodologies, and they covered various aspects of gastroenterology, including patient education, self-assessment, patient–physician communication, disease management, and research question generation. The flowchart delineating the selection procedure of the studies included is depicted in Figure 1 .

Flowchart delineating the selection procedure of the studies included in the review.

Table 1 summarizes the characteristics of studies included in the review.

Characteristics of studies included in the review.

GERD, gastroesophageal reflux disease; GI, gastrointestinal; GPT, generative pre-trained transformer.

Table 2 summarizes the main findings, and the identified benefits and limitations of ChatGPT in gastroenterology as presented in the studies included.

Key findings and limitations of ChatGPT as identified by studies included.

AI, artificial intelligence; GERD, gastroesophageal reflux disease; PPI, proton pump inhibitor.

Table 3 summarizes the quality assessment performed according to JBI critical appraisal tools.

Quality assessment of studies included in the review using modified JBI critical appraisal tools.

JBI, Joanna Briggs Institute.

ChatGPT as a tool for patients

Two studies examined the efficacy of ChatGPT as a tool for patients, mainly in answering common patient questions. The first study by Lee et al. , 6 ‘ChatGPT answers common patient questions about colonoscopy’ found that ChatGPT provided answers similar in quality to non-AI answers but the text similarity was low. The AI-generated answers were written at higher grade reading levels, exceeding the recommended eighth-grade threshold.

In another study, 9 our group evaluated the utility of ChatGPT in answering a variety of clinical questions addressing a wide range of topics, including common symptoms, diagnostic tests, and treatments for various gastrointestinal conditions. The study revealed that ChatGPT could provide accurate and clear answers in some cases, but not in others, indicating the need for further development. Notably, both studies only examined GPT-3.5, an older and less capable ChatGPT model that is free to access.

ChatGPT as a tool for physicians

As a tool for physicians, ChatGPT was evaluated in several aspects: clinical reasoning, knowledge, and education.

In the clinical field, ChatGPT was evaluated in two different domains; management of Gastroesophageal Reflux Disease (GERD), 11 and optimizing post-colonoscopy management. 12

In the study ‘Evaluation of the potential utility of an artificial intelligence ChatBot in GERD management’, 11 evaluated the utility of ChatGPT in the management of GERD. The authors did not specify which ChatGPT model they investigated for this study. The results showed that ChatGPT provided appropriate and specific recommendations for GERD management in 91.3% of cases, with 29.0% considered completely appropriate and 62.3% mostly appropriate. However, inconsistencies were noted in responses to the same prompt, and some potential proton pump inhibitor (PPI) risks were stated as facts. Notably, patients from diverse educational backgrounds universally regarded the responses as comprehensible and beneficial. Furthermore, all respondents expressed their inclination to consider the tool as a valuable resource for obtaining medical information, highlighting the superior utility of the response format compared to that of a conventional search engine.

In the second domain, Gorelik et al. 12 evaluated the responses generated by ChatGPT to 20 clinical scenarios presented to ChatGPT in the form of structured reports and free-text notes. The responses generated by ChatGPT were assessed focusing on measuring adherence to guidelines, accuracy of responses, and inter-rater agreement between the gastroenterologists. Results demonstrated a 90% adherence to guidelines and an 85% accuracy in responses.

In the subject of knowledge and education, Suchman et al. 7 evaluated the performance of ChatGPT on the American College of Gastroenterology Self-Assessment Tests.

Surprisingly, ChatGPT was not able to pass multiple versions of the exams, indicating its limitations with gastroenterology subspeciality level question and answer tasks. Notably, in this study, the authors examined and compared both versions of ChatGPT (versions GPT-3.5 and GPT-4), with no actual difference in the results achieved: ChatGPT-3.5 scored 65.1% and GPT-4 scored 62.4% (passing grade was 70%).

ChatGPT as a tool for researchers

In assessing ChatGPT as a tool for researchers, our group evaluated the use of a ChatGPT for highlighting research priorities and identifying open and meaningful top research questions in gastroenterology. 8 The research questions outputted by ChatGPT achieved high ratings for relevance and clarity as well as an average rating for specificity but performed poorly in terms of originality.

Studies heterogeneity

When diving into the complexities of research involving AI models, understanding the underlying methods is pivotal. Several factors can introduce variability in the outcomes of such studies, especially when dealing with models like ChatGPT. Factors contributing to this heterogeneity include the following:

• The absence of specific model version details (e.g. 3.5 versus 4.0 or the exact date of each model). Different versions may exhibit varied performances.
• Lack of clarity on the method of question submission – whether they were sent collectively or in individual chat sessions for each prompt and response. The chosen method can significantly influence the replies.
• Non-disclosure of the prompts utilized in the research, hindering the reproducibility of the findings.
• Uncertainty over whether the study repeated the same prompts to gauge the consistency in the model’s responses. The criteria for prompt selection remain vague. Even minor alterations in the wording can drastically change outcomes, a phenomenon particularly evident with GPT 3.5. Enhancing the quality of prompts can substantially refine many of these investigations.

Benefits and limitations of ChatGPT in gastroenterology

From the included studies, several benefits of ChatGPT in gastroenterology were identified. These include the ability of ChatGPT to provide appropriate and specific recommendations, aid in patient–physician communication, patient education, and continuous patient care, and generate relevant and clear research questions. However, limitations were also noted, including ChatGPT’s insufficient understanding of complex medical questions, inconsistencies in responses, some potential PPI risks being stated as fact, some responses providing limited specific guidance, and struggles with originality. Ethical considerations were also raised, such as confidentiality and data security, stereotypes, bias and inaccuracy, plagiarism concerns, compliance with data privacy regulations, and the irreplaceable role of human judgment.

The results of this systematic review provide a comprehensive evaluation of the role of ChatGPT, an LLM, in the field of gastroenterology. The included studies highlight the potential of ChatGPT in various applications, including patient education, self-assessment, patient–physician communication, disease management, and research question generation. However, they also underscore several limitations and ethical considerations that warrant further exploration and careful regulation.

In the era of digital health, it is essential to critically evaluate emerging technologies and their potential impact on healthcare delivery. This review contributes to this ongoing discourse, offering a focused examination of ChatGPT in the context of gastroenterology.

Our review focused on ChatGPT, as it stands out as the most popular LLM chat tool, for patients and physicians alike. Therefore, it is important to assess its performance on tasks relevant to both groups.

We believe that the insights gleaned from this review will be valuable not only to practitioners and researchers in gastroenterology, but also to policymakers, AI developers, and the broader healthcare community as we navigate the integration of AI into healthcare. 4 , 5

The studies evaluating the efficacy of ChatGPT in answering common patient questions 6 , 9 reveal a mixed picture. While ChatGPT demonstrated the ability to generate credible medical information, its performance was inconsistent. Some responses were accurate and clear, while others were not, indicating an insufficient understanding of complex medical information. Moreover, the AI-generated answers were written at significantly higher grade reading levels than recommended, potentially limiting their accessibility to patients with lower literacy levels. Notably, this point can easily be fixed with prompt adjustments at a system level. These findings echo the cautionary note in the editorial ‘Will ChatGPT transform healthcare?’, 4 and highlight the need for further development and fine-tuning of ChatGPT to ensure its reliability and accessibility in patient education.

In the context of gastroenterology board exam-style medical reasoning, ChatGPT did not achieve a passing score using the methods in the recently published study. This indicates its limitations as an educational tool in its current form. 7 Notably, this study examined and compared the performance of both versions of the chatbot – the free version (GPT-3.5) and the advanced version (ChatGPT-4). It is noteworthy that the advanced version did not demonstrate an advantage over the free version. On the contrary, it lagged by three points. This finding emphasizes the need for continuous updates and the development of fine-tuned models specifically geared toward medical education, as suggested in the study, or the use of additional LLM augmentation methods like database linkage. Given the dynamic nature of medical knowledge, AI tools used in medical education need to be capable of providing accurate and updated information and following new information and guidelines that become available.

ChatGPT shows promise in enhancing patient–physician communication and continuous patient care. 10 Its ability to take patients’ medical history, present the information in a concise, structured format, and continuously learn and improve based on the responses it receives could potentially improve healthcare outcomes. Moreover, by taking on tasks such as patient education and medical history taking, ChatGPT could help reduce physician burnout. However, the ethical considerations and limitations of AI, including confidentiality and data security, stereotypes, bias and inaccuracy, plagiarism concerns, compliance with data privacy regulations, and the irreplaceable role of human judgment, need to be addressed. AI technologies like ChatGPT should complement, and not replace, the human elements of empathy and professional judgment.

In the management of GERD, ChatGPT provided appropriate and specific recommendations in the majority of cases. 11 However, inconsistencies in responses to the same prompt and some potential PPI risks being stated as fact were identified as limitations. These findings highlight the need for rigorous clinical oversight in the use of ChatGPT in disease management.

ChatGPT’s ability to generate relevant, clear, and moderately specific research questions is noteworthy. 8 However, it struggled with originality, suggesting the need for further work to improve the novelty of the generated research questions.

While it is not disclosed what was the specific training data used to train ChatGPT, it is likely that similar to other LLMs, it was trained on vast information derived from the internet and other open-access sources. However, ChatGPT is not innately attuned to medical nuances. 6 , 9 This may explain its inconsistency in providing clear and accurate information on gastroenterological issues.

In addition, ChatGPT’s information might not always be up-to-date. Especially, regarding recent medical research and guidelines. 7 Its training data depends on available open-source information at the time of training. Thus, it might not be aware of newer studies or guidelines unless it is retrained on newer data.

A significant challenge is ChatGPT’s language complexity. 6 This complexity is not an inherent flaw but a byproduct of the data it was trained on. However, for patient interactions, it is beneficial that the information is delivered at an accessible reading level.

Bias and the potential for manipulation through prompt engineering 6 arise because the model reflects the data it was trained on. If biases exist in those datasets, they will also be present in the model’s outputs. This can be hazardous in medical applications where impartiality is essential.

The inconsistencies in responses to similar prompts 11 may be a byproduct of the model’s vast training data. This is because the model may attempt to generate varied responses. However, in a clinical context, consistency is vital. Thus, this unpredictable behavior is a clear limitation.

A limitation demonstrated in the field of gastroenterology, particularly in medical reasoning and board exams, is the lack of domain-specific training for ChatGPT. 7 A model specifically fine-tuned on gastroenterological data could potentially outperform the general ChatGPT.

Notably, this could be related to how the task was presented to the model and a limitation of ChatGPT in particular. Perhaps with prompt engineering and increasing the temperature (creativity) of the model, more original responses could have been created.

This finding aligns with the review published by Sharma and Parasa, 3 which emphasizes the need for careful implementation and regulation of AI tools in healthcare.

ChatGPT proved its capability in handling various scenarios and descriptions effectively, providing concise patient letters in post-colonoscopy management by offering guideline-based recommendations. 12 These findings suggest that ChatGPT has the potential to assist healthcare providers in streamlining post-colonoscopy decision-making and improving adherence to post-colonoscopy surveillance guidelines.

This review has some limitations. First, the number of studies included in the review was relatively small, limiting the generalizability of the findings. However, as far as we know, it summarizes the current literature on the topic of ChatGPT in the field of gastroenterology. Second, the included studies were diverse in their objectives and methodologies, making it challenging to make quantitative analyses.

Finally, given the fast-paced advancements in AI technology, the conclusions of this review might soon be obsolete. Notably, the majority of the studies assessed the free version of ChatGPT. 3 , 5 However, recent research suggests that the improved version, ChatGPT-4, performs better in the medical field. 13

In conclusion, the review of ChatGPT’s application in gastroenterology reveals mixed outcomes. While showing promise as a tool for physicians (e.g. in GERD management and post-colonoscopy adherence to guidelines), it struggled with inconsistencies in patient education and failed in self-assessment tests. However, most data were created using the free version ChatGPT, 3 , 5 while the improved version (GPT-4) may achieve better results. Our findings emphasize the potential of ChatGPT and also underline clear limitations and the need for further refinement and ethical scrutiny.

Acknowledgments

Contributor Information

Eyal Klang, Division of Data-Driven and Digital Medicine (D3M), Icahn School of Medicine at Mount Sinai, New York, NY, USA. The Charles Bronfman Institute of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA. ARC Innovation Center, Sheba Medical Center at Tel Hashomer Affiliated with Tel Aviv Medical School, Tel Aviv University, Tel Aviv, Israel.

Ali Sourosh, Division of Data-Driven and Digital Medicine (D3M), Icahn School of Medicine at Mount Sinai, New York, NY, USA. The Charles Bronfman Institute of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA.

Girish N. Nadkarni, Division of Data-Driven and Digital Medicine (D3M), Icahn School of Medicine at Mount Sinai, New York, NY, USA. The Charles Bronfman Institute of Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA.

Kassem Sharif, Department of Gastroenterology, Sheba Medical Center at Tel Hashomer Affiliated with Tel Aviv Medical School, Tel Aviv University, Tel Aviv, Israel.

Adi Lahat, Department of Gastroenterology, Sheba Medical Center at Tel Hashomer, Ramat Gan, 52621 Affiliated with Tel Aviv Medical School, Tel Aviv University, Tel Aviv, Israel.

Declarations

Ethics approval and consent to participate: Not applicable.

Consent for publication: All authors consented publication.

Author contributions: Eyal Klang: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Software; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing.

Girish N. Nadkarni: Investigation; Supervision; Writing – original draft; Writing – review & editing.

Kassem Sharif: Investigation; Resources; Validation; Visualization; Writing – original draft; Writing – review & editing.

Adi Lahat: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Writing – original draft; Writing – review & editing.

Funding: The authors received no financial support for the research, authorship, and/or publication of this article.

The authors declare that there is no conflict of interest.

Availability of data and materials: Not applicable.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• 18 January 2023

ChatGPT listed as author on research papers: many scientists disapprove

• Chris Stokel-Walker 0

Chris Stokel-Walker is a freelance journalist in Newcastle, UK.

You can also search for this author in PubMed   Google Scholar

The artificial-intelligence (AI) chatbot ChatGPT that has taken the world by storm has made its formal debut in the scientific literature — racking up at least four authorship credits on published papers and preprints.

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 51 print issues and online access

185,98 € per year

only 3,65 € per issue

Rent or buy this article

Prices vary by article type

Prices may be subject to local taxes which are calculated during checkout

Nature 613 , 620-621 (2023)

doi: https://doi.org/10.1038/d41586-023-00107-z

Kung, T. H. et al. Preprint at medRxiv https://doi.org/10.1101/2022.12.19.22283643 (2022).

O’Connor, S. & ChatGPT Nurse Educ. Pract. 66 , 103537 (2023).

Article   PubMed   Google Scholar  

ChatGPT & Zhavoronkov, A. Oncoscience 9 , 82–84 (2022).

GPT, Osmanovic Thunström, A. & Steingrimsson, S. Preprint at HAL https://hal.science/hal-03701250 (2022).

Download references

Reprints and permissions

Related Articles

Are ChatGPT and AlphaCode going to replace programmers?

The immune system can sabotage gene therapies — can scientists rein it in?

News 28 MAY 24

Guidelines for academics aim to lessen ethical pitfalls in generative-AI use

Nature Index 22 MAY 24

Are robots the solution to the crisis in older-person care?

Outlook 25 APR 24

Who will make AlphaFold3 open source? Scientists race to crack AI model

News 23 MAY 24

Egypt is building a $1-billion mega-museum. Will it bring Egyptology home?

News Feature 22 MAY 24

Pay researchers to spot errors in published papers

World View 21 MAY 24

Postdoctoral Associate - Amyloid Strain Differences in Alzheimer's Disease

Houston, Texas (US)

Baylor College of Medicine (BCM)

Postdoctoral Associate- Bioinformatics of Alzheimer's disease

Postdoctoral associate- alzheimer's gene therapy, postdoctoral associate, kaust global postdoctoral fellowship.

The KAUST Global Fellowship Program is designed to attract emerging research leaders working across areas under the four research priorities of KAUST.

Saudi Arabia (SA)

King Abdullah University of Science and Technology (KAUST

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

'ZDNET Recommends': What exactly does it mean?

ZDNET's recommendations are based on many hours of testing, research, and comparison shopping. We gather data from the best available sources, including vendor and retailer listings as well as other relevant and independent reviews sites. And we pore over customer reviews to find out what matters to real people who already own and use the products and services we’re assessing.

When you click through from our site to a retailer and buy a product or service, we may earn affiliate commissions. This helps support our work, but does not affect what we cover or how, and it does not affect the price you pay. Neither ZDNET nor the author are compensated for these independent reviews. Indeed, we follow strict guidelines that ensure our editorial content is never influenced by advertisers.

ZDNET's editorial team writes on behalf of you, our reader. Our goal is to deliver the most accurate information and the most knowledgeable advice possible in order to help you make smarter buying decisions on tech gear and a wide array of products and services. Our editors thoroughly review and fact-check every article to ensure that our content meets the highest standards. If we have made an error or published misleading information, we will correct or clarify the article. If you see inaccuracies in our content, please report the mistake via this form .

What is ChatGPT and why does it matter? Here's what you need to know

What is ChatGPT?

ChatGPT is an AI chatbot with natural language processing (NLP) that allows you to have human-like conversations to complete various tasks. The  generative AI  tool can answer questions and assist you with tasks such as composing emails, essays, code, and more.

Also :  How to use ChatGPT: What you need to know now

It's currently  open to use for free . A paid subscription version called ChatGPT Plus launched in February 2023 with access to priority access to OpenAI's latest models and updates.

Who made ChatGPT?

AI startup OpenAI launched ChatGPT on November 30, 2022. OpenAI has also developed  DALL-E 2  and DALL-E 3 , popular  AI image generators , and Whisper, an automatic speech recognition system. 

Who owns ChatGPT currently?

OpenAI owns ChatGPT. Microsoft is a major investor in OpenAI thanks to multiyear, multi-billion dollar  investments. Elon Musk was an investor when OpenAI was first founded in 2015, but has since completely severed ties with the startup and created his own AI chatbot, Grok .

How can you access ChatGPT?

On April 1, 2024, OpenAI stopped requiring you to log in to use ChatGPT. Now, you can access ChatGPT simply by visiting  chat.openai.com . You can also access ChatGPT  via an app on your iPhone  or  Android  device.

Once you visit the site, you can start chatting away with ChatGPT. A great way to get started is by asking a question, similar to what you would do with Google. You can ask as many questions as you'd like.

Also: ChatGPT no longer requires a login, but you might want one anyway. Here's why

There are still some perks to creating an OpenAI account, such saving and reviewing your chat history and accessing custom instructions. Creating an OpenAI account is entirely free and easy. You can even log in with your Google account.

For step-by-step instructions, check out ZDNET's guide on  how to start using ChatGPT . 

Is there a ChatGPT app?

Yes, an official ChatGPT app is available for both iPhone and Android users. 

Also: ChatGPT dropped a free app for iPhones. Does it live up to the hype?

Make sure to download OpenAI's app, as there are a plethora of copycat fake apps listed on Apple's App Store and the Google Play Store that are not affiliated with the startup.

Is ChatGPT available for free?

ChatGPT is free to use, regardless of what you use it for, including writing, coding, and much more. 

There is a subscription option , ChatGPT Plus, that users can take advantage of that costs $20/month. The paid subscription model guarantees users extra perks, such as priority access to GPT-4o and the latest upgrades. 

Also: ChatGPT vs ChatGPT Plus: Is it worth the subscription fee?

Although the subscription price may seem steep, it is the same amount as Microsoft Copilot Pro and Google One AI, Microsoft's and Google's premium AI offerings. 

The free version is still a solid option as it can access the same model and most of the same perks. One major exception: only subscribers get guaranteed access to GPT-4o when the model is at capacity. 

I tried using ChatGPT and it says it's at capacity. What does that mean?

The ChatGPT website operates using servers. When too many people hop onto these servers, they may overload and can't process your request. If this happens to you, you can visit the site later when fewer people are trying to access the tool. You can also keep the tab open and refresh it periodically. 

Also: The best AI chatbots

If you want to skip the wait and have reliable access, you can subscribe to  ChatGPT Plus  for general access during peak times, faster response times, and priority access to new features and improvements, including priority access to GPT-4o.

You can also try using Bing's AI chatbot, Copilot . This chatbot is free to use, runs on GPT-4, has no wait times, and can access the internet for more accurate information.

What is ChatGPT used for?

ChatGPT has many functions in addition to answering simple questions. ChatGPT can compose essays , have philosophical conversations, do math, and even code for you . 

The tasks ChatGPT can help with also don't have to be so ambitious. For example, my favorite use of ChatGPT is for help creating basic lists for chores, such as packing and grocery shopping, and to-do lists that make my daily life more productive. The possibilities are endless. 

ZDNET has published many ChatGPT how-to guides. Below are some of the most popular ones. 

Use ChatGPT to: 

• Write an essay
• Create an app
• Build your resume
• Write Excel formulas
• Summarize content
• Write a cover letter
• Start an Etsy business
• Create charts and tables
• Write Adruino drivers

Can ChatGPT generate images?

Yes, ChatGPT can generate images, but only for ChatGPT Plus subscribers. Since OpenAI discontinued DALL-E 2 in February 2024, the only way to access its most advanced AI image generator, DALL-E 3, through OpenAI's offerings is via its chatbot and ChatGPT Plus subscription.

Also: DALL-E adds new ways to edit and create AI-generated images. Learn how to use it

Microsoft's Copilot offers image generation, which is also powered by DALL-E 3, in its chatbot for free. This is a great alternative if you don't want to shell out the money for ChatGPT Plus.

How does ChatGPT work?

ChatGPT runs on a large language model (LLM) architecture created by OpenAI called the  Generative Pre-trained Transformer  (GPT). Since its launch, the free version of ChatGPT ran on a fine-tuned model in the GPT-3.5 series until May 2024, when the startup upgraded the model to GPT-4o. 

Also:   Here's a deep dive into how ChatGPT works  

With a subscription to ChatGPT Plus , you can access GPT-3.5, GPT-4, or  GPT-4o . Plus, users also have the added perk of priority access to GPT-4o, even when it is at capacity, while free users get booted down to GPT-3.5. 

Generative AI models of this type are trained on vast amounts of information from the internet, including websites, books, news articles, and more.

What does ChatGPT stand for?

As mentioned above, the last three letters in ChatGPT's namesake stand for Generative Pre-trained Transformer (GPT), a family of large language models created by OpenAI that uses deep learning to generate human-like, conversational text. 

Also: What does GPT stand for? Understanding GPT 3.5, GPT 4, GPT-4 Turbo, and more

The "Chat" part of the name is simply a callout to its chatting capabilities. 

Is ChatGPT better than a search engine?

ChatGPT is a language model created to converse with the end user. A search engine indexes web pages on the internet to help users find information. One is not better than the other, as each suit different purposes. 

When searching for as much up-to-date, accurate information as you can access, your best bet is a search engine. It will provide you with pages upon pages of sources you can peruse. 

Also: The best AI search engines of 2024: Google, Perplexity, and more

As of May, the free version of ChatGPT can get responses from both the GPT-4o model and the web. It will only pull its answer from, and ultimately list, a handful of sources, as opposed to showing nearly endless search results.

For example, I used GPT-4o to answer, "What is the weather today in San Francisco?" The response told me it searched four sites and provided links to them. 

If you are looking for a platform that can explain complex topics in an easy-to-understand manner, then ChatGPT might be what you want. If you want the best of both worlds, there are plenty of AI search engines on the market that combine both.

What are ChatGPT's limitations?

Despite its impressive capabilities, ChatGPT still has limitations. Users sometimes need to reword questions multiple times for ChatGPT to understand their intent. A bigger limitation is a lack of quality in responses, which can sometimes be plausible-sounding but are verbose or make no practical sense. 

Instead of asking for clarification on ambiguous questions, the model guesses what your question means, which can lead to poor responses. Generative AI models are also subject to hallucinations, which can result in inaccurate responses.

Does ChatGPT give wrong answers?

As mentioned above, ChatGPT, like all language models, has  limitations  and can give nonsensical answers and incorrect information, so it's important to double-check the data it gives you.

Also: 8 ways to reduce ChatGPT hallucinations

OpenAI recommends that you provide feedback on what ChatGPT generates by using the thumbs-up and thumbs-down buttons to improve its underlying model. You can even join the startup's Bug Bounty program , which offers up to $20,000 for reporting security bugs and safety issues.

Can ChatGPT refuse to answer my prompts?

AI systems like ChatGPT can and do reject  inappropriate requests . The AI assistant can identify inappropriate submissions to prevent the generation of unsafe content.

Also:  6 things ChatGPT can't do (and another 20 it refuses to do)

These submissions include questions that violate someone's rights, are offensive, are discriminatory, or involve illegal activities. The ChatGPT model can also challenge incorrect premises, answer follow-up questions, and even admit mistakes when you point them out.

These guardrails are important. AI models can generate advanced, realistic content that can be exploited by bad actors for harm, such as spreading misinformation about public figures and influencing elections .

Can I chat with ChatGPT?

Although some people use ChatGPT for elaborate functions, such as writing code or even malware , you can use ChatGPT for more mundane activities, such as having a friendly conversation. 

Also:  Do you like asking ChatGPT questions? You could get paid (a lot) for it

Some conversation starters could be as simple as, "I am hungry, what food should I get?" or as elaborate as, "What do you think happens in the afterlife?" Either way, ChatGPT is sure to have an answer for you. 

Is ChatGPT safe?

People are expressing concerns about AI chatbots replacing or atrophying human intelligence. For example, a chatbot can write an article on any topic efficiently (though not necessarily accurately) within seconds, potentially eliminating the need for human writers.

Chatbots can also write an entire essay within seconds, making it easier for students to cheat or avoid learning how to write properly. This even led  some school districts to block access  when ChatGPT initially launched. 

Also:  Generative AI can be the academic assistant an underserved student needs

Now, not only have many of those schools decided to unblock the technology, but some higher education institutions have been  catering their academic offerings  to AI-related coursework. 

Another concern with AI chatbots is the possible spread of misinformation. ChatGPT itself says: "My responses are not intended to be taken as fact, and I always encourage people to verify any information they receive from me or any other source." OpenAI also notes that ChatGPT sometimes writes "plausible-sounding but incorrect or nonsensical answers."

Also:  Microsoft and OpenAI detect and disrupt nation-state cyber threats that use AI, report shows

Lastly, there are ethical concerns regarding the information ChatGPT was trained on, since the startup scraped the internet to train the chatbot. 

It also automatically uses people's interactions with the free version of the chatbot to further train its models, raising privacy concerns. OpenAI lets you turn off training in ChatGPT's settings.

Does ChatGPT plagiarize?

Yes, sort of. OpenAI scraped the internet to train ChatGPT. Therefore, the technology's knowledge is influenced by other people's work. Since there is no guarantee that when OpenAI outputs its answers it is entirely original, the chatbot may regurgitate someone else's work in your answer, which is considered plagiarism. 

Is there a ChatGPT detector?

Concerns about students using AI to cheat mean the need for a ChatGPT text detector is becoming more evident. 

In January 2023, OpenAI released a free tool to target this problem. Unfortunately, OpenAI's "classifier" tool could only correctly identify 26% of AI-written text with a "likely AI-written" designation. Furthermore, it provided false positives 9% of the time, incorrectly identifying human-written work as AI-produced. 

The tool performed so poorly  that, six months after being released, OpenAI it shut down "due to its low rate of accuracy." Despite the tool's failure, the startup claims to be researching more effective techniques for AI text identification.

Also: OpenAI unveils text-to-video model and the results are astonishing

Other AI detectors exist on the market, including GPT-2 Output Detector ,  Writer AI Content Detector , and Content at Scale's AI Content Detection  tool. ZDNET put these tools to the test, and the results were underwhelming: all three were found to be unreliable sources for spotting AI, repeatedly giving false negatives. Here are  ZDNET's full test results .

What are the common signs something was written by ChatGPT?

Although tools aren't sufficient for detecting ChatGPT-generated writing, a  study  shows that humans could detect AI-written text by looking for politeness. The study's results indicate that  ChatGPT's writing style is extremely polite . And unlike humans, it cannot produce responses that include metaphors, irony, or sarcasm.

Will my conversations with ChatGPT be used for training?

One of the major risks when using generative AI models is that they become more intelligent by being trained on user inputs. Therefore, when familiarizing yourself with how to use ChatGPT, you might wonder if your specific conversations will be used for training and, if so, who can view your chats.

Also:  This ChatGPT update fixed one of my biggest productivity issues with the AI chatbot

OpenAI will use your conversations with the free chatbot to automatically training data to refine its models. You can opt out of the startup using your data for model training by clicking on the question mark in the bottom left-hand corner, Settings, and turning off "Improve the model for everyone."

What is GPT-4?

GPT-4 is OpenAI's language model that is much more advanced than its predecessor, GPT-3.5. Users can access GPT-4 by subscribing to ChatGPT Plus for $20 per month or using Microsoft's Copilot.

Also: What does GPT stand for? Understanding GPT-3.5, GPT-4, and more

GPT-4 has advanced intellectual capabilities, meaning it outperforms GPT-3.5 in a series of simulated benchmark exams. The newer model also supposedly produces fewer hallucinations. 

What is GPT-4o?

GPT-4o is OpenAI's latest, fastest, and most advanced flagship model. As the name implies, it has the same intelligence as GPT-4. However, the "o" in the title stands for "omni," referring to its multimodal capabilities, which allow it to understand text, audio, image, and video inputs and output text, audio, and image outputs. 

Also:  6 ways OpenAI just supercharged ChatGPT for free users

The model is 50% cheaper in the API than GPT-4 Turbo while still matching its English and coding capabilities and outperforming it in non-English languages, vision, and audio understanding -- a big win for developers.

Are there alternatives to ChatGPT worth considering?

Although ChatGPT gets the most buzz, other options are just as good -- and might even be better suited to your needs. ZDNET has created a list of the best chatbots, which have all been tested by us and show which tool is best for your requirements. 

Also: 4 things Claude AI can do that ChatGPT can't

Despite ChatGPT's extensive abilities, there are major downsides to the AI chatbot. If you want to try the technology, there are plenty of other options: Copilot , Claude , Perplexity ,  Jasper , and more.  

Is ChatGPT smart enough to pass benchmark exams?

Yes, ChatGPT is capable of passing a series of benchmark exams. A professor at Wharton, the University of Pennsylvania's business school, used ChatGPT to take an MBA exam and the results were quite impressive. 

ChatGPT not only passed the exam, but the tool scored between a B- and a B. The professor, Christian Terwiesch, was impressed at its basic operations management, process analysis questions, and explanations.

OpenAI also tested the chatbot's ability to pass benchmark exams. Although ChatGPT could pass many of these benchmark exams, its scores were usually in the lower percentile. However, with GPT-4, ChatGPT can score much higher.

For example, ChatGPT using GPT-3.5 scored in the lower 10th percentile of a simulated Bar Exam, while GPT-4 scored in the top 10th percentile. You can see more examples from OpenAI in the chart below.

Can ChatGPT be used for job application assistance?

Yes, ChatGPT is a great resource to help with job applications. Undertaking a job search can be tedious and difficult, and ChatGPT can help you lighten the load. ChatGPT can build your resume  and write a cover letter .

Also :  How to use ChatGPT to write an essay

If your application has any written supplements, you can use ChatGPT to help you write those essays or personal statements . 

What are the most common ChatGPT plugins, and how do I use them?

Plugins allowed ChatGPT to connect to third-party applications, including access to real-time information on the web. The plugins expanded ChatGPT's abilities , allowing it to assist with many more activities, such as planning a trip or finding a place to eat. 

Also:  My two favorite ChatGPT Plus features and the remarkable things I can do with them

On March 19, 2024, however, OpenAI stopped allowing users to install new plugins or start new conversations with existing ones. Instead, OpenAI replaced plugins with GPTs , which are easier for developers to build. 

Users can find 3 million ChatGPT chatbots, also known as GPTs, on the GPT store. Unfortunately, there is also a lot of spam in the GPT store.

What is Microsoft's involvement with ChatGPT?

Microsoft was an early investor in OpenAI, the AI startup behind ChatGPT, long before ChatGPT was released to the public. Microsoft's first involvement with OpenAI was in 2019, when the company invested $1 billion, and then another $2 billion in the years after. In January 2023, Microsoft extended its partnership with OpenAI through a multiyear, multi-billion dollar investment .

Also: ChatGPT vs. Copilot: Which AI chatbot is better for you?

 Neither company disclosed the investment value, but unnamed sources told Bloomberg that it could total $10 billion over multiple years. In return, OpenAI's exclusive cloud-computing provider is Microsoft Azure, powering all OpenAI workloads across research, products, and API services.

Microsoft has also used its OpenAI partnership to revamp its Bing search engine and improve its browser. 

On February 7, 2023, Microsoft unveiled a new Bing tool , now known as Copilot, that runs on OpenAI's GPT-4, customized specifically for search.

What does Copilot (formerly Bing Chat) have to do with ChatGPT?

In February 2023,  Microsoft unveiled  a new version of Bing -- and its standout feature was its integration with ChatGPT. When it was announced, Microsoft shared that Bing Chat, now Copilot, was powered by a next-generation version of OpenAI's large language model, making it "more powerful than ChatGPT." Five weeks after the launch, Microsoft revealed that Copilot had been running on GPT-4 before the model had even launched. 

How does Copilot compare to ChatGPT?

Copilot uses OpenAI's GPT-4, which means that since its launch, it has been more efficient and capable than the standard, free version of ChatGPT. At the time, Copilot boasted several other features over ChatGPT, such as access to the internet, knowledge of current information, and footnotes.

In May 2024, however, OpenAI supercharged the free version of its chatbot with GPT-4o. The upgrade gave users GPT-4 level intelligence, the ability to get responses from the web via ChatGPT Browse with Bing, analyze data, chat about photos and documents, use GPTs, access the GPT Store, and Voice Mode. Therefore, after the upgrade, ChatGPT reclaimed its crown as the best AI chatbot. 

What is Gemini and how does it relate to ChatGPT?

Gemini is Google's AI chat service, a rival to ChatGPT. On February 6, 2023, Google introduced its experimental AI chat service, which was then called Google Bard. Over a month after the announcement, Google began rolling out  access to Bard first via a waitlist . Now, it is available to the general public. 

Artificial Intelligence

Chatgpt vs. copilot: which ai chatbot is better for you, how to use chatgpt (and how to access gpt-4o), what does gpt stand for understanding gpt-3.5, gpt-4, gpt-4o, and more.

Technology and policy in India

Manipur High Court Uses ChatGPT To Conduct Research During Verdict

We missed this earlier: The Manipur High Court, during a judgement on 23 May, 2024, used ChatGPT to conduct research before passing an order, according to a report by Live Law. The case pertained to petitioner Md. Zakir Hussain, a retired member of the Village Defence Force of his district in Manipur. Hussain was dismissed from service in 2021 without receiving a copy of his dismissal order, due to a criminal having escaped from the police station he was working at while he was on duty.

Subsequently, in December 2023, Justice A Guneshwar Sharma, the presiding judge of the Manipur High Court, sought a response from the police counsel, Advocate Shyam Sharma, on the grounds for such a dismissal.

However, when he discovered the affidavit submitted by the police to be inadequate, he referred to GPT 3.5 to conduct further legal research.

What Was The Final Judgement?

GPT 3.5 provided Justice Guneshwar with information pertaining to Manipur’s Village Defence Force (VDF), to the effect that it consists of volunteers from local communities, who are trained in assisting the police in dealing with insurgency and intercommunity conflicts in order to improve local security in rural areas of the state. Based on this information, and further investigations conducted by the Court which revealed that a Show Cause notice needs to be issued to dismissed personnel to explain any charges against them, Hussain was reinstated to his service.

How Have Indian Courts Used ChatGPT previously?

In a previous Medianama article , we had covered the launch of the SUPACE (Supreme Court Portal for Assistance in Courts Efficiency) technology, based on Artificial Intelligence, which was inaugurated by the then Chief Justice SA Bobde. Justice L Nageshwara Rao, the Chairperson of the Supreme Court’s Artificial Intelligence Committee, had emphasized that AI would not make lawyers and judges redundant as it would only be used for routine administrative tasks, but could be a “tremendous asset” to the workings of the Court. Current CJI Justice DY Chandrachud has also spoken about the importance of AI and the digitization of court proceedings.

On the one hand, Indian judges have been using ChatGPT for research purposes. For instance, in March 2023, Justice Anoop Chitkara of the Punjab and Haryana High Court used ChatGPT in a bail hearing, involving accused Jaswinder Singh who had assaulted an individual leading to his death. Justice Chitkara used the AI Model to seek information about bail jurisprudence in cases involving ‘cruelty’ while committing a homicide and was given information to the effect that bail can be denied in such cases.

In other cases, lawyers have been reprimanded by courts for relying on ChatGPT and other AI Language Models. In August 2023, the Delhi High Court observed in a trademark dispute , involving designer Christian Louboutin, that GPT cannot be used by lawyers to provide reasoning on ‘legal or factual matters in a court of law’, as stated by Justice Pratibha Singh.

Singh was wary of how ChatGPT and AI Language Models could be used to generate “incorrect information, imaginative data or fictional laws”, a trend which has been witnessed in other countries as well.

Why This Matters

The fact that AI generated content, although fictitious, could appear believable in a court of law, is troubling. Further, while the United Kingdom, among other countries, has guidelines restricting the use of AI Language Models in courts, India lacks such regulations for the use of generative AI and platforms such as ChatGPT in court proceedings.

This could potentially lead to problems such as misinformation and over-reliance on AI in judicial proceedings.

• “Legal determinations often involve gray areas that still require application of human judgment”: US Chief Justice Writes on AI in Courtrooms
• Supreme Court launches AI-based system to help judges, CJI assures it won’t spill over to decision making
• US Lawyer uses ChatGPT for case research, ends up citing fake cases in legal brief

STAY ON TOP OF TECH NEWS:  Our daily newsletter with the top story of the day from MediaNama, delivered to your inbox before 9 AM. Click  here  to sign up today!

OpenAI Offers Nonprofits Discounts on Corporate ChatGPT Product

OpenAI logo is seen in this illustration taken March 31, 2023. REUTERS/Dado Ruvic/Illustration

By Anna Tong

SAN FRANCISCO (Reuters) - - OpenAI will offer discounts on corporate ChatGPT subscriptions to nonprofit organizations, the Microsoft-backed firm said on Thursday, as it seeks to grow sales of its artificial intelligence product to enterprises.

Under the program, large nonprofits will be able to get 50% off the enterprise-grade version of ChatGPT. Smaller nonprofits using ChatGPT Team will pay $20 per month per user, instead of $25 or $30.

When OpenAI released the consumer-focused ChatGPT in November 2022, it set off frenzied use of generative AI in daily tasks from writing to coding and became the fastest application to acquire 100 million users.

Recently, the company has courted large organizations to buy its ChatGPT enterprise product, which it plans to make a bigger part of its revenue stream.

The company also announced on Thursday a partnership with the International Rescue Committee (IRC).

OpenAI has granted $250,000 to the IRC, a New York-based nonprofit that helps those affected by humanitarian crises, such as the war in Ukraine.

The funds will go towards developing an education chatbot built on OpenAI technology to assist educators in crisis zones.

Called aprendIA, it is currently being piloted in places such as Bangladesh and Nigeria, the IRC said.

"A total of 224 million school-aged and pre-school aged children are affected by crises globally...in conflict and crisis settings, teachers lack resources to support children who face unique challenges," the IRC said in a statement.

IRC said educators will be able to use aprendIA to create interactive and personalized teaching content adaptable to different humanitarian contexts.

To expand its global reach, IRC plans to make chatbots available on low-tech devices and within existing chat products like WhatsApp and Facebook Messenger in multiple languages.

(Reporting by Anna Tong in San Francisco; Editing by Sam Holmes)

Copyright 2024 Thomson Reuters .

Join the Conversation

Tags: artificial intelligence , United States

America 2024

Health News Bulletin

Stay informed on the latest news on health and COVID-19 from the editors at U.S. News & World Report.

Sign in to manage your newsletters »

Sign up to receive the latest updates from U.S News & World Report and our trusted partners and sponsors. By clicking submit, you are agreeing to our Terms and Conditions & Privacy Policy .

You May Also Like

The 10 worst presidents.

U.S. News Staff Feb. 23, 2024

Cartoons on President Donald Trump

Feb. 1, 2017, at 1:24 p.m.

Photos: Obama Behind the Scenes

April 8, 2022

Photos: Who Supports Joe Biden?

March 11, 2020

Trump Guilty in Historic Trial

Lauren Camera May 30, 2024

What to Know: Human Cases of Bird Flu

Cecelia Smith-Schoenwalder May 30, 2024

How’s Biden Doing With Young Voters?

Q1 Economic Growth Revised Downward

Tim Smart May 30, 2024

When Will the Israel-Hamas War End?

Cecelia Smith-Schoenwalder May 29, 2024

Biden Tries to Energize Black Voters

Laura Mannweiler May 29, 2024