ARTICLE

 

ChatGPT in Political Education: A Comparative Study of Student and Faculty Perspectives

 

 

Hao Song^{I, II}

^IInstitute of International Relations, Nanjing University, China
^IIShanghai Lixin University of Accounting and Finance, China. 2114974631@qq.com. https://orcid.org/0009-0000-5306-6685

 

 

---

ABSTRACT

This mixed-methods study explores the credibility of responses generated by artificial intelligence (AI) in facilitating the learning of international politics by comparing students' and faculty's perspectives on using ChatGPT to learn international politics. A total of 207 students and five faculty members were recruited in this study. Results from sentiment analysis indicate that students usually trust ChatGPT as a credible source of information. They also consider it an effective tool for understanding complex political concepts. However, findings of thematic analysis show that faculty highlight significant limitations of using ChatGPT for learning international politics. These experts discuss a lack of depth in AI-generated responses, potential ideological biases, and insufficient critical analysis of nuanced issues. These differences indicate the need for cautious integration of AI tools in learning international politics. Overall, the findings suggest that while ChatGPT can serve as a valuable supplementary resource, it should not replace traditional learning methods or expert guidance.

Keywords: artificial intelligence (AI); ChatGPT; international politics; higher education

---

 

 

Introduction

Artificial intelligence (AI) refers to "a system that displays intelligent behavior by analyzing the environment and taking a degree of autonomous action to achieve a specific objective" (European Commission 2018, 1). One widely used AI tool is ChatGPT, which is a conversational AI system that interactively generates human-like responses. This AI tool has a wide range of applications in education, research, marketing, software engineering, and healthcare (Fraiwan and Khasawneh 2023; Sallam 2023). Consequently, ChatGPT has gained popularity for producing content and obtaining information.

Given its ability to rapidly generate responses to user queries based on the vast amount of information it was trained on, ChatGPT offers a wide array of possibilities for both experts and the general public across various sectors of society (Azaria, Azoulay, and Reches 2024; Haque and Li 2025). For individuals without specialised knowledge, ChatGPT can be a valuable resource for gaining insights into certain topics, deciphering technical terms, or receiving guidance on personal projects (Gruda 2024). By making information more accessible, ChatGPT democratises knowledge, allowing people to engage with subjects that may otherwise be difficult to understand. This can lower the barriers to learning and empowers individuals to participate in discussions that were previously the domain of experts (Shahzad, Xu, and Zahid 2025). As a result, this tool contributes to the broader popularisation and dissemination of knowledge, fostering an informed and inclusive society (Shabbir et al. 2024). When using it for learning, students may perceive ChatGPT as an expert in their disciplines, making it an ideal "know-it-all" (Han et al. 2024, 2).

AI systems are built using human-created algorithms and user-generated data. Therefore, they inevitably reflect human biases, social norms, and political ideologies (Huang 2025). AI inherits cognitive biases from both users and designers who generate biases related to gender, ethnicity, religion, and politics (Motoki, Neto, and Rodrigues 2024). However, compared to other kinds of algorithmic bias, the subject of political prejudice in AI systems has received less attention. Messer (2025) notes that AI may generate politically biased content. If ChatGPT generates replies that are prejudiced in the area of national politics, such information may impact users' perceptions and judgements of the global context. This study aims to explore whether solely depending on ChatGPT to seek answers may influence learners' perspectives of understanding international politics. This study uses China's artificial intelligence cooperation with the Association of Southeast Asian Nations (China-ASEAN AI cooperation) as an example of an international political topic to investigate this question. The research questions include:

1. What are students' perspectives on ChatGPT-generated answers for learning topics related to international politics?

2. What are faculty's perspectives on ChatGPT-generated answers for learning topics related to international politics?

3. What are the differences in perspectives on ChatGPT-generated answers for learning topics related to international politics between students and faculty?

 

Literature Review

Using ChatGPT in Education

AI has been prevalently used in the field of education. According to Loeckx (2016), AI holds the potential to reduce workloads for educators and students, provide engaging learning experiences, and drive educational trends such as resource digitalisation, gamification, and personalised learning.

Among the many AI tools, ChatGPT is most commonly used in education. Its ability to synthesise large volumes of text and generate content has greatly contributed to its growing popularity in teaching and learning environments (Baidoo-Anu and Ansah 2023). By providing quick, concise, and coherent responses, ChatGPT can serve as a valuable tool for students and educators; its capacity to break down complex ideas, provide instant feedback, and engage in interactive learning enhances understanding across various subjects, leading to its widespread adoption in educational settings (Memarian and Doleck 2023). Students usually view ChatGPT as an on-demand tutor that helps them understand learning materials, seek feedback, and brainstorm ideas (Baillifard et al. 2025). However, educators worry about the overreliance on using AI for learning, which may influence students' critical thinking (Gerlich 2025; Zhai, Wibowo, and Li 2024).

Today, ChatGPT is being used widely in education, and people are starting to focus on how this AI tool impacts political subject learning. Using ChatGPT to learn political knowledge has the potential to help students understand political concepts, political systems, and how they work. For example, Garg and Garg (2023) studied how Indian students used ChatGPT for learning, and the results showed that students who used this tool understood key political concepts better than those who did not. This study also looked at how ChatGPT helped with discussions on contemporary political issues, showing that participants felt more confident and could express their opinions more clearly when discussing political topics. Ardoin and Hicks (2024) also mention that ChatGPT can be a helpful tool for students to learn political knowledge, and using AI may be beneficial for their future careers in campaigning, government, or law. However, using AI also comes with ethical issues and challenges.

Challenges of Using ChatGPT in Learning

Using ChatGPT in educational practice has raised several potential issues, such as communication barriers, limited understanding, biased training data, lack of creative thinking, insufficient context understanding, and privacy risks (Baidoo-Anu and Ansah 2023). A global survey conducted by Van Noorden and Perkel (2023) among over 1,600 researchers found that when asked about the possible negative impacts of generative AI, 68% of respondents expressed concerns regarding the spread of false information, warning that AI could be used for fraud that is hard to detect. Additionally, 66% of respondents worried that incorrect or misleading content might appear in academic articles, reflecting a growing concern about the challenges that AI presents in educational environments.

Likewise, Han et al. (2024) evaluated how ChatGPT can be used for medical education. They discovered that although ChatGPT can serve as a valuable resource, it occasionally provides inaccurate responses to fundamental scientific inquiries as well as misleading medical information. Similarly, Casey (2024) conducted an assessment task in which 25 students used ChatGPT to draft a policy brief for an Australian government minister. This study then examined these students' perspectives towards using ChatGPT for public policy development and education. Results from students' self-reflective essays and focus groups reveal that ChatGPT often fails to produce analytically sound and nuanced policy recommendations. The findings corresponded with the technoscepticism theoretical framework, emphasising concerns that AI tools may undermine effective policy analysis procedures (Newman and Mintrom 2023).

Additionally, ChatGPT may bring broader social and political risks. For instance, Akgun and Greenhow (2022) note that using AI in education could strengthen systemic biases and discrimination, further increasing inequality among historically marginalised student groups. They also state that AI-generated information could even amplify various forms of bias, such as racism, sexism, and xenophobia. In other words, inherent bias may exist in the pre-training data, which could cause AI systems such as ChatGPT to develop political-ideological biases. Liu et al. (2023) emphasise that we must be especially cautious when deploying AI tools in public areas such as education and advertising, as these biases could significantly influence opinions and outcomes. This concern needs to be paid attention because digital literacy plays a key role in shaping individuals' judgements on political issues (King 2019). In the context of international politics, the bias in AI-generated information may lead to flawed prediction models, which may fail to accurately identify major geopolitical events or correctly assess their trends and impacts. Therefore, when using AI to learn international politics, which usually involve sensitive and crucial topics, careful and unbiased examination is a must.

Credibility for Learning Purpose Model

Scholars (Fogg 2003; Fogg and Tseng 1999) mention two key dimensions of credibility of online resources or messages: trustworthiness and expertise. Trustworthiness is defined as being well-intentioned and unbiased, and expertise refers to perceived knowledge, ability, and experience. Accordingly, this study was guided by a credibility for learning purpose model adopted from previous studies (see Figure 1). Perceived credibility for using ChatGPT to learn is evaluated through examining two key dimensions: trustworthiness and expertise of ChatGPT's responses. Trustworthiness evaluates whether ChatGPT-generated content is perceived as reliable and unbiased, and expertise measures the perceived knowledge including the depth of the information delivered in the content. By focusing on these dimensions, this study explores how effective students and faculty consider ChatGPT's ability for learning international politics.

 

 

Methods

Activity Design and Data Collection

Participants read ChatGPT-generated answers on four sets of questions related to the ASEAN-China AI cooperation. Because the complexity of international issues involves multiple variables and dynamics, each set of questions contains a macro-level question with its answer, as well as a micro-level question with its answer. Macro issues focus on the structure of the international system, the balance of power among states, and global governance, all of which influence state relations and policies. On the other hand, micro issues focus on the interactions between individual states and non-state actors, encompassing policymaking, diplomatic strategies, and public opinion. By integrating both macro and micro perspectives, we could effectively understand the intricacies of international issues and uncover potential linkages and trends.

A total of eight questions (see Table 1) were developed for ChatGPT to answer. The design of the questions was generated by three faculty members who are experts in the field of international politics, ensuring they were thoughtfully developed rather than randomly generated, thereby enhancing the credibility of the research approach.

 

 

Students read ChatGPT-generated answers and then shared their thoughts to evaluate the credibility of the ChatGPT-generated content. A total of 207 out of 210 (response rate equals 98.6%) undergraduate students from a range of academic disciplines, including economics, law, business administration, computer science, foreign languages and literature, and international relations across two research institutions in East China were recruited. First-year students accounted for a smaller proportion, with 21 individuals (10.1%), while second-year and third-year students formed the majority of the participants with 105 (50.7%) and 81 (39.2%), respectively. Moreover, 117 (56.5%) participants were female students and 90 (43.5%) were male students. Students completed an anonymous questionnaire that was distributed via the online survey platform SoJump, and the survey remained open for two weeks.

One-on-one semi-structured interviews with five faculty members as experts in the field of international politics were conducted to further understand the credibility of ChatGPT-generated information related to ASEAN-China AI cooperation. They have teaching experience in the field ranging from 11 to 29 years (see Table 2). These experts read ChatGPT-generated answers and then shared their perspectives on the answers, along with using ChatGPT for learning international politics.

 

 

Instruments

Students' perspectives on the credibility of ChatGPT-generated answers were examined through three questions adopted from previous literature (Nov, Singh, and Mann 2023), including "The presented answers can be a more trustworthy alternative to Baidu or other search engines to answer my questions about topics related to China's Artificial Intelligence Cooperation with ASEAN", "The presented answers can be a more trusted alternative to professional learning (e.g., taking relevant courses, joining professional seminars or reports) for answering my questions on ASEAN-related topics", and "The presented answers could help me better understand topics related to China's Artificial Intelligence Cooperation with ASEAN." Meanwhile, the one-on-one interviews explored experts' perspectives on the same ChatGPT-generated answers. Each interview lasted for about 40 minutes. Sample questions include, "As a teacher, would you consider using ChatGPT's answers to help students understand this topic?"

Data Analysis

This study used a mixed-methods approach to explore the research questions. First, sentiment analysis was conducted on 207 student responses regarding the credibility of ChatGPT-generated answers. The Bing lexicon was used to classify and evaluate the sentiment of each response by assigning positive or negative values to words, enabling the calculation of an overall sentiment score by subtracting the total negative values from the positive values for each response. Sentiment analysis was chosen in this study because it provides an efficient and systematic way to capture the underlying attitudes in students' responses, thereby offering a quantifiable measure of their perceptions beyond surface-level content (Medhat, Hassan, and Korashy 2014).

To supplement the quantitative results, a multiple-case study was then conducted to explore key themes from the interviews with the five faculty members. This method worked well for looking at the "how" and "why" of trusting ChatGPT-generated information for learning because it focused on current events and did not need much intervention from researchers to control the data (Yin 2014). Each participant was viewed as a unique case, and thematic analysis was conducted to evaluate the interview responses following Braun and Clarke's (2006) guidelines. The initial coding process involved utilising participants' responses to develop a codebook. In the second step, the author worked to find the main topics using an open-coding method that was inductive. In the last phase, the themes were fine-tuned to ensure robust and useful results (Thornberg and Charmaz 2014; Yin 2014). Together, both quantitative and qualitative results provide a deeper understanding of participants' perspectives, allowing for richer insights.

Findings

Sentiment analysis was conducted on the question concerning whether the AI-generated answers can be viewed as a more trusted alternative to Baidu or other search engines. Figure 2 shows the contribution of individual words to the overall sentiment. "Comprehensive" received the strongest positive sentiment and accounts for 24% among words with more than two frequencies. Other words, such as "clear" (11%), "logical" (12%), and "accurate" (5.2%), contributed moderately positive sentiment, while "problems" (3.4%) and "bad" (2.6%) received the strongest negative sentiment. This suggests a generally positive perception.

 

 

Figure 3 illustrates the distribution of overall sentiment scores. Each response was evaluated by calculating the sentiment score, derived by subtracting the total negative value from the positive value. A prominent peak is observed around a sentiment score of one (positive), with smaller peaks in the less positive and negative ranges. This pattern confirms the predominantly positive general assessment presented in Figure 2.

 

 

The word cloud in Figure 4 visually reinforces the findings. "Comprehensive" is the largest word, highlighting its importance in positive sentiment. Negative words ("problems", "bad") are present but smaller than many other positive terms.

 

 

Second, regarding the question of whether AI-generated answers can be a more trusted alternative to professional learning, Figure 5 shows that "enough" (20.4%), "comprehensive" (15.7%), and "clear" (7.4%) receive strong positive sentiment. However, "lack" (12%) is a prominent negative term. This result suggests that some respondents found AI-generated answers could provide enough professionalism, while others believed these answers lacked certain aspects when compared to professional learning opportunities.

 

 

Figure 6 somewhat skewed towards the positive side (scores around positive 1-2), but a notable portion of responses falls closer to neutral or slightly negative, which aligns with the presence of "lack" in Figure 5. This indicates that students are somewhat sceptical towards using ΑΙ as an alternative for professional learning.

 

 

The world cloud in Figure 7 shows that "comprehensive" and "enough" are the most prominent words, suggesting perceived sufficiency and completeness, but "lack" is still visible, reflecting the concerns raised in using AI as an alternative to professional learning.

 

 

Lastly, when asked whether the AI-generated answers can help them better understand ASEAN-related topics, Figure 8 shows that "comprehensive" (20.7%) is again the top contributor to positive sentiment, followed by "better [understand]" (14%) and "clear" (9.8%), indicating its continued importance, while "lack" (3.3%) shows up as a negative term.

 

 

Figure 9 furthermore indicates that the distribution is highly skewed towards positive sentiment, with a massive peak around a score of one. This shows that responses were overwhelmingly positive towards using AI-generated answers to understand ASEAN-related topics.

 

 

Lastly, the word cloud in Figure 10 supports the strong positive sentiment, with "comprehensive" being prominent. "Better [understanding]" also stands out, showcasing its strength in positive expression.

 

 

Overall, "comprehensive" consistently receives the most positive sentiment, indicating that the comprehensiveness of the answers generated by AI is a key strength. In contrast, terms such as "lack" and "problems" suggest areas where improvement is needed to reach the desired level of trustworthiness and usefulness for users compared to established alternatives such as professional learning. These words may reflect students' scepticism that was often related to ChatGPT's inability to provide accurate content.

Next, thematic analysis of experts' reflections showed three themes: 1) Lack of depth in ChatGPT's responses, 2) Use of ChatGPT for learning, and 3) Use of ChatGPT to learn international politics for general public.

Theme One: Lack of Depth in ChatGPT's Responses

The experts generally agreed that ChatGPT's responses fall short in terms of depth, particularly when it comes to addressing complex aspects of international politics. As one expert explained, "International politics often involves complex and multifaceted issues that are not easily distilled into simple answers, making it difficult for AI to capture the full complexity and interconnected nature of these matters." For instance, one faculty member pointed out that "[a]ll four sets of answers lack scientific thinking. Even though the answers are rich, they are still shallow because they only point out a few main points without going into much more details." Another faculty member shared the following:

For example, the historical relationship between two countries can significantly influence their current political interactions. As a result, ChatGPT's responses to these four sets of questions lack analysis of historical and cultural background factors and do not provide in-depth exploration of individual arguments. The answers, therefore, fall short in depth.

This statement was similar to the opinion shared by another faculty member, who noted that "AI may also struggle to provide the nuanced historical context necessary to fully understand the current dynamics of international politics".

Additionally, one expert expressed that "ChatGPT does have limitations regarding region- and country-specific questions". This faculty member furthermore said:

First, it may not fully catch the nuances of local culture, customary expressions, or the specific context of a region, leading to answers that can seem general or overlook important subtleties. Second, for highly specialised or technical issues related to a particular region or country, ChatGPT may lack the depth of expertise or detailed local knowledge that regional experts provide. Third, it may not have up-to-date information, especially concerning fast-moving topics such as political events, economic changes, or public health crises. The availability and quality of data can vary significantly from region to region, impacting the accuracy and reliability of responses. For instance, areas with fewer records may yield less accurate information.

Lastly, other faculty agreed that ChatGPT-generated answers have a certain degree of credibility, but the depth of analysis is insufficient, because "[t]he responses remain at the level of objective description, lacking professionalism and theoretical rigour, and often exhibit a homogeneous expression across multiple questions". In short, these critiques highlight the limitations of AI-generated responses, which require a deep and context-rich understanding of international political topics.

Theme Two: Use of ChatGPT for Learning

Generally, the faculty considered ChatGPT as a valuable tool for providing basic knowledge, which can be useful for helping students grasp fundamental concepts and themes related to international politics. For example, one faculty mentioned, "I would consider using ChatGPT to assist students in learning about topics related to international politics. For non-majors, ChatGPT can help them grasp basic concepts and themes in international politics." This perspective highlights ChatGPT's potential to facilitate an initial understanding of complex subjects, offering descriptions or preliminary analyses that can serve as a foundation for further learning.

However, the participating experts expressed concerns about ChatGPT's limitations regarding deeper learning needs. They noted that this AI tool lacks the ability to deliver the depth and critical analysis that are required for more comprehensive understanding. For instance, one faculty member mentioned, "for students who are majored in international relations and related fields, the depth of ChatGPT answers is insufficient". Another faculty echoed this concern: "Its responses are primarily descriptive and may lack the depth and critical analysis necessary for a comprehensive understanding." These observations underline the inadequacy of ChatGPT in addressing the complexities of advanced topics, particularly those that require an in-depth approach.

Furthermore, some experts highlighted the unique challenges posed by the subject of international politics, where the complexity of issues often exceeds ChatGPT's capabilities. As one said,

Due to the unique nature of international relations, characterised by the anarchy of the international community, this assistance is limited to the understanding of concepts and descriptions of events. It falls short of interpreting the complexities of international political issues at a deeper level.

This comment suggests that while ChatGPT can be a helpful supplementary resource, it cannot replace the need for more rigorous and comprehensive materials and discussions.

These experts additionally shared their concerns about the broader implications of using ChatGPT to learn international politics. One faculty member warned that the targeted and algorithmic nature of AI information could significantly influence young students' worldviews and values, potentially impacting ideological security and national political stability: "The limitations of information dissemination shaped by the recommendation mechanisms of perspective algorithms are still difficult to overcome." This faculty member further suggested that "when using ChatGPT to help students understand and learn about international political issues, it is essential to prioritise ideological security and address political bias to prevent harmful political ideas from influencing students' judgements on international matters". These concerns emphasise the importance of carefully considering how ChatGPT should be integrated into educational settings, particularly when dealing with sensitive and complex subjects.

Theme Three: The Use of ChatGPT for the General Public to Learn International Politics

The faculty members agreed that ChatGPT could aid the general public to obtain a preliminary understanding of international politics. One expert praised the tool and believed that it "is an excellent learning aid and is adept at answering questions, sometimes even providing perfect responses". Another faculty member echoed this sentiment, noting that "responses [by ChatGPT] are comprehensive and can serve as a valuable resource for non-international politics majors and the general public to understand international political issues". Some faculty members furthermore discussed the benefits of using ChatGPT as a learning tool, emphasising that it can "help the general public better comprehend and participate in discussions on international relations". They also concluded three key advantages: "the ability to ask questions and receive instant feedback, which deepens understanding, ... the stimulation of interest in international affairs, which encourages critical thinking and deeper discussions, ... and the provision of comprehensive answers that aid in understanding international political events".

However, several limitations were highlighted. One is that this AI tool "falls short of interpreting the complexities of international political issues at a deeper level". Another significant issue is the ethical considerations of using ChatGPT. As one expert explained:

I would use ChatGPT as a supplementary teaching aid, given the limitations of its question-and-answer mode of communication. Additionally, due to the constraints in corpus selection, data cleaning, and algorithm design, dialogue-based education may face various ethical and moral risks. Therefore, I will advise students to be mindful of these issues while using the tool.

Bias in ChatGPT's responses was another concern. The experts stressed that ChatGPT-generated answers may not be as accurate or fair because of its potential ideological bias. One noted,

It is essential to remain aware of the risks associated with ideological biases. A technological hegemon may exploit large models to promote ideological preferences, "weaponising" ideology to exploit the psychological vulnerabilities of other countries and engage in subtle ideological penetration and institutional subversion, thereby threatening national security.

They further suggested that when encouraging people to use ChatGPT for exploring related issues, "it is crucial to recognise the information preferences within its database. This awareness can help mitigate any biases in ChatGPT's understanding of certain topics and assist individuals in gaining a clearer understanding of relevant issues."

Language differences are another concern when using ChatGPT to learn international politics. One faculty member highlighted this: "The same question may yield different answers in different languages. For example, responses to the same question can vary between English and Chinese." He additionally noted that this would as well lead to ideological algorithmic bias. Cultural differences were meanwhile identified as another important factor to be considered because "[t]he content of information bases or corpora can vary significantly across different communities. For instance, while prejudice against black people may be prevalent in the United States, it may be less pronounced in China, leading to different understandings of black rights."

 

Discussions

To answer the research questions, the findings show that students perceive ChatGPT as a credible tool for learning international politics. However, the faculty hold more reserved opinions (see Figure 11).

 

 

In answering the first research question, generally, students believed that ChatGPT's responses were credible in terms of answering questions about ASEAN-China AI cooperation. Students expressed that ChatGPT could be a credible alternative to other search engines such as Baidu. Furthermore, when learning about ASEAN-China AI cooperation, some students believed ChatGPT could serve as a reliable alternative to professional learning (e.g., taking a relevant course or attending a seminar), while some doubted ChatGPT's professionalism regarding using it to replace professional learning.

Finally, students hold a strong belief that ChatGPT could help them better understand topics related to ASEAN-China AI cooperation.

In answering the second and third research questions, compared to students, the faculty members showed a lower trust towards ChatGPT's ability to help learners understand international politics, mirroring previous conclusions that students usually view ChatGPT as credible, while experts focus more on its limitations (Fogg and Tseng 1999). Specifically, the faculty experts highlighted ChatGPT's limitations in addressing complicated international political issues, including its superficial theoretical depth, absence of critical analysis, susceptibility to ideological prejudice, and its failure to comprehensively grasp the nuances of more sophisticated topics. These findings suggest that although ChatGPT may be beneficial for fundamental learning, it may not be sufficient for in-depth learning.

Furthermore, some faculty members expressed concerns that the underlying data and algorithms may reflect ideological perspectives about international political issues as perceived by non-experts. This finding echoes previous conclusions that the bias of ChatGPT-generated information poses a significant challenge for educational systems (Kasneci et al. 2023). Moreover, the dissemination of biased content related to international politics may result in institutional bias, functioning in a manner that benefits certain social groups at the expense of others, and altering social power structures and relationships (Peters 2022). Some faculty members additionally warned that in the international community, where governance is decentralised, disparities in AI technology among countries may also create a digital divide between technologically advanced and undeveloped nations. These concerns revealed that if technologically advanced countries use the AI tools that they developed to spread ideologically biased international political information to less technologically developed countries, this may impact the perspectives and values of young people in the countries that are not technologically advanced. Such influence on the educational content and worldview of these students may pose risks to national security by changing the ideological landscape and potentially destabilising existing social and political structures. Therefore, these concerns at some level call for cautious use of AI tools for learning about complex and sensitive fields.

Overall, ChatGPT should not be considered as a replacement for professional learning in international politics because ChatGPT-generated answers often lack arguments and rationalisations, even when references are cited. ChatGPT-generated information may also contain ideological biases. The algorithmic techniques underlying AI may influence users' understanding of political issues through the algorithm's handling of attitudes, perceptions, and expressions regarding geopolitical issues, thus potentially distorting public discourse and undermining informed decision-making (Huang 2025). Therefore, it is vital to involve human experts to check and regulate AI's answers.

In order to effectively use ChatGPT for learning international politics, instructors should encourage students to critically assess ChatGPT-generated information instead of simply accepting the answers, thus enhancing their critical thinking and analytical skills. Moreover, ChatGPT can be used as a starting point to stimulate curiosity, guide initial exploration, and provide basic understanding. Meanwhile, the instructors should ask students to compare AI-generated content with other sources. This would help them identify and understand international political topics from different approaches. Lastly, the instructors should integrate diverse and credible sources into the curriculum to supplement AI-generated information. By including credible evidence and examples, including those from non-Western or marginalised voices, a balanced view regarding international political topics can be provided. This strategy would ensure that students access multiple viewpoints and identify biases that may be present in AI-generated information.

 

Conclusions

Several limitations existed. First, this study explores one case in international politics (i.e., China-ASEAN AI cooperation) focusing on four aspects. Future research should involve broader international political cases as well as expand aspects to gain a more comprehensive understanding of using ChatGPT for learning international politics. Second, one-on-one interviews should be conducted to provide further insights into students' perceptions of using ChatGPT for learning international politics. Furthermore, there is a need to compare AI-generated and human-generated responses to analyse students' and faculty's perspectives on different types of feedback for in-depth understanding. Moreover, researchers should look at the long-term impact of using AI tools for learning on students' critical thinking. Finally, with the rapid development of AI technology, there is also a need to develop adaptable teaching strategies to embrace the technology while still maintaining academic rigour and intellectual integrity.

In conclusion, this study investigates both students' and faculty's perspectives on using ChatGPT for learning international politics. Our findings show that ChatGPT has the potential to enhance learning in international politics. However, its limitations call for a careful and strategic approach when using it. By emphasising ethical considerations, motivating critical thinking, and integrating diverse perspectives, educators and learners could effectively use AI tools as useful resources in the educational landscape.

 

Funding

This research received grants from the Ministry of Education (MOE) of China, under the Humanities and Social Sciences Foundation (Project number: 23JDSZ3106).

 

References

Akgun, S., and C. Greenhow. 2022. "Artificial Intelligence in Education: Addressing Ethical Challenges in K-12 Settings". AI and Ethics 2 (3): 431-440. https://doi.org/10.1007/s43681-021-00096-7.         [ Links ]

Ardoin, P. J., and W. D. Hicks. 2024. "Fear and Loathing: ChatGPT in the Political Science Classroom". PS: Political Science and Politics 57 (4): 583-593. https://doi.org/10.1017/s1049096524000131.         [ Links ]

Azaria, A., R. Azoulay, and S. Reches. 2024. "ChatGPT Is a Remarkable Tool-For Experts". Data Intelligence 6 (1): 240-296. https://doi.org/10.1162/dint_a_00235.         [ Links ]

Baidoo-Anu, D., and O. L. Ansah. 2023. "Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning". Journal of AI 7 (1): 52-62. https://doi.org/10.2139/ssrn.4337484.

Baillifard, A., M. Gabella, P. B. Lavenex, and C. S. Martarelli. 2025. "Effective Learning with a Personal AI Tutor: A Case Study". Education and Information Technologies 30 (1): 297-312. https://doi.org/10.1007/s10639-024-12888-5.         [ Links ]

Braun, V., and V. Clarke. 2006. "Using Thematic Analysis in Psychology". Qualitative Research in Psychology 3 (2): 77-101. https://doi.org/10.1191/1478088706qp063oa.         [ Links ]

Casey, D. 2024. "ChatGPT in Public Policy Teaching and Assessment: An Examination of Opportunities and Challenges". Australian Journal of Public Administration 1-15. https://doi.org/10.1111/1467-8500.12647.

European Commission. 2018. "Artificial Intelligence for Europe". COM/2018/237 Final. Accessed November 8, 2025. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=COM:2018:237:FIN.

Fogg, B. J. 2003. Persuasive Technology: Using Computers to Change What We Think and Do. San Francisco: Morgan Kaufmann. https://doi.org/10.1145/764008.763957.         [ Links ]

Fogg, B. J., and H. Tseng. 1999. "The Elements of Computer Credibility". In CHI '99: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 80-87. New York: Association for Computing Machinery. https://doi.org/10.1145/302979.303001.

Fraiwan, M., and N. Khasawneh. 2023. "A Review of ChatGPT Applications in Education, Marketing, Software Engineering, and Healthcare: Benefits, Drawbacks, and Research Directions". Preprint, arXiv, April 29. https://doi.org/10.48550/arXiv.2305.00237.

Garg, J., and K. Garg. 2023. "ChatGPT as an Empowering Catalyst: Unveiling the Impact on Political Awareness and Civic Education". i-manager's Journal on School Educational Technology 19 (2): 1-12. https://doi.org/10.26634/jsch.19.2.20149.         [ Links ]

Gerlich, M. 2025. "AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking". Societies 15 (1): 6. https://doi.org/10.3390/soc15010006.         [ Links ]

Gruda, D. 2024. "Three Ways ChatGPT Helps Me in My Academic Writing". Nature, April 8. https://doi.org/10.1038/d41586-024-01042-3.

Han, Z., F. Battaglia, A. Udaiyar, A. Fooks, and S. R. Terlecky. 2024. "An Explorative Assessment of ChatGPT as an Aid in Medical Education: Use It with Caution". Medical Teacher 46 (5): 657-664. https://doi.org/10.1080/0142159X.2023.2271159.         [ Links ]

Haque, M. A., and S. Li. 2025. "Exploring ChatGPT and Its Impact on Society". AI and Ethics 5: 791-803. https://doi.org/10.1007/s43681-024-00435-4.         [ Links ]

Huang, Z. A. 2025. "Terminology, AI Bias, and the Risks of Current Digital Public Diplomacy Practices". Place Branding and Public Diplomacy 21: 327-333. https://doi.org/10.1057/s41254-024-00324-x.         [ Links ]

Kasneci, E., K. Sessler, S. Küchemann, M. Bannert, D. Dementieva, F. Fischer, U. Gasser, G. Groh, S. Günnemann, E. Hüllermeier, S. Kusche, G. Kutyniok, T. Michaeli, C. Nerdel, J. Pfeffer, O. Poquet, M. Sailer, A. Schmidt, T. Seidel, M. Stadler, and G. Kasneci. 2023. "ChatGPT for Good? On Opportunities and Challenges of Large Language Models for Education". Learning and Individual Differences 103: 102274. https://doi.org/10.1016/j.lindif.2023.102274.         [ Links ]

King, K. 2019. "Education, Digital Literacy and Democracy: The Case of Britain's Proposed 'Exit' from the European Union (Brexit)". Asia Pacific Education Review 20: 285-294. https://doi.org/10.1007/s12564-019-09594-0.         [ Links ]

Liu, Y., T. Han, S. Ma, J. Zhang, Y. Yang, J. Tian, H. He, A. Li, M. He, Z. Liu, Z. Wu, L. Zhao, D. Zhu, X. Li, N. Qiang, D. Shen, T. Liu, and B. Ge. 2023. "Summary of ChatGPT-Related Research and Perspectives towards the Future of Large Language Models". Meta-Radiology 1 (2): 100017. https://doi.org/10.1016/j.metrad.2023.100017.         [ Links ]

Loeckx. 2016. "Blurring Boundaries in Education: Context and Impact of MOOCs". The International Review of Research in Open and Distributed Learning 17 (3): 92-121. https://doi.org/10.19173/irrodl.v17i3.2395.

Medhat, W., A. Hassan, and H. Korashy. 2014. "Sentiment Analysis Algorithms and Applications: A Survey". Ain Shams Engineering Journal 5 (4): 1093-1113. https://doi.org/10.1016/j.asej.2014.04.011.         [ Links ]

Memarian, B., and T. Doleck. 2023. "ChatGPT in Education: Methods, Potentials and Limitations". Computers in Human Behavior: Artificial Humans 1 (2): 100022. https://doi.org/10.1016/j.chbah.2023.100022.         [ Links ]

Messer, U. 2025. "How Do People React to Political Bias in Generative Artificial Intelligence (AI)?" Computers in Human Behavior: Artificial Humans 3: 100108. https://doi.org/10.1016/j.chbah.2024.100108.         [ Links ]

Motoki, F., V. P. Neto, and V. Rodrigues. 2024. "More Human Than Human: Measuring ChatGPT Political Bias". Public Choice 198 (1): 3-23. https://doi.org/10.1007/s11127-023-01042-8.         [ Links ]

Newman, J., and M. Mintrom. 2023. "Mapping the Discourse on Evidence-Based Policy, Artificial Intelligence, and the Ethical Practice of Policy Analysis". Journal of European Public Policy 30 (9): 1839-1859. https://doi.org/10.1080/13501763.2023.2193223.         [ Links ]

Nov, O., N. Singh, and D. Mann. 2023. "Putting ChatGPT's Medical Advice to the (Turing) Test: Survey Study". JMIR Medical Education 9: e46939. https://doi.org/10.2196/46939.         [ Links ]

Peters, U. 2022. "Algorithmic Political Bias in Artificial Intelligence Systems". Philosophy and Technology 35 (2): 25. https://doi.org/10.1007/s13347-022-00512-8.         [ Links ]

Sallam, M. 2023. "ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid Concerns". Healthcare 11 (6): 887. https://doi.org/10.3390/healthcare11060887.         [ Links ]

Shabbir, A., S. F. Rizvi, M. M. Alam, and M. M. Su'ud. 2024. "Beyond Boundaries: Navigating the Positive Potential of ChatGPT, Empowering Education in Underdeveloped Corners of the World". Heliyon: A Cell Press Journal 10 (16): e35845. https://doi.org/10.1016/j.heliyon.2024.e35845.         [ Links ]

Shahzad, M. F., S. Xu, and H. Zahid. 2025. "Exploring the Impact of Generative AI-Based Technologies on Learning Performance through Self-Efficacy, Fairness and Ethics, Creativity, and Trust in Higher Education". Education and Information Technologies 30: 3691-3716. https://doi.org/10.1007/s10639-024-12949-9.         [ Links ]

Thornberg, R., and K. Charmaz. 2014. "Grounded Theory and Theoretical Coding". In The SAGE Handbook of Qualitative Data Analysis, edited by U. Flick, 153-169. London: Sage Publications. https://doi.org/10.4135/9781446282243.n11.

Van Noorden, R., and J. M. Perkel. 2023. "AI and Science: What 1,600 Researchers Think". Nature 621 (7980): 672-675. https://doi.org/10.1038/d41586-023-02980-0.         [ Links ]

Yin, R. K. 2014. Case Study Research: Design and Methods. 5th ed. Thousand Oaks: Sage Publications.         [ Links ]

Zhai, C., S. Wibowo, and L. D. Li. 2024. "The Effects of Over-Reliance on AI Dialogue Systems on Students' Cognitive Abilities: A Systematic Review". Smart Learning Environments 11 (1): 28. https://doi.org/10.1186/s40561-024-00316-7.         [ Links ]