Actionable companion · Assessment & reasoning

From AI evidence to responsible practice: Higher education students’ perceptions of ChatGPT: A global study of early reactions

AI-assisted editorial implementation guide, separate from the original research. Proposed actions are not findings or recommendations attributed to the source authors. No independent peer review of this guide is asserted.

Read original paper Search paper agent CC BY 4.0

Evidence and context

Title and abstract screening; stage labels describe reported settings, not verified numerical ages. Design labels require full-text appraisal. Reported stage(s): Higher education. Verify the participants, intervention and comparator before applying this study in a new setting.

Original authors’ abstract, reproduced under the linked source license:

Abstract The paper presents the most comprehensive and large-scale global study to date on how higher education students perceived the use of ChatGPT in early 2024. With a sample of 23,218 students from 109 countries and territories, the study reveals that students primarily used ChatGPT for brainstorming, summarizing texts, and finding research articles, with a few using it for professional and creative writing. They found it useful for simplifying complex information and summarizing content, but less reliable for providing information and supporting classroom learning, though some considered its information clearer than that from peers and teachers. Moreover, students agreed on the need for AI regulations at all levels due to concerns about ChatGPT promoting cheating, plagiarism, and social isolation. However, they believed ChatGPT could potentially enhance their access to knowledge and improve their learning experience, study efficiency, and chances of achieving good grades. While ChatGPT was perceived as effective in potentially improving AI literacy, digital communication, and content creation skills, it was less useful for interpersonal communication, decision-making, numeracy, native language proficiency, and the development of critical thinking skills. Students also felt that ChatGPT would boost demand for AI-related skills and facilitate remote work without significantly impacting unemployment. Emotionally, students mostly felt positive using ChatGPT, with curiosity and calmness being the most common emotions. Further examinations reveal variations in students’ perceptions across different socio-demographic and geographic factors, with key factors influencing students’ use of ChatGPT also being identified. Higher education institutions’ managers and teachers may benefit from these findings while formulating the curricula and instructions/regulations for ChatGPT use, as well as when designing the teaching methods and assessment tools. Moreover, policymakers may also consider the findings when formulating strategies for secondary and higher education system development, especially in light of changing labor market needs and related digital skills development.

What kind of evidence is this?

Supporting AI learning science · Design requires full-text appraisal

Stage: Higher education. Screening labels describe reported context and study-design language. They are not a risk-of-bias rating or proof of effectiveness for an age band.

Study methods · source passage

Our global study targeted higher education students who are currently enrolled at any level in a higher education institution, are at least 18 years old, and have the legal capacity to provide free and voluntary consent to participate in an anonymous online survey. Survey participants were recruited using a convenience sampling method, which involved promoting the survey in classrooms and through advertisements on university communication systems. This method proved effective due to its practical nature, allowing for easy access to potential participants who were readily available and willing to participate in the survey. The survey was initially designed in English. Due to the lack of systematic evidence on students’ early perceptions of ChatGPT, the content of the questionnaire was developed in collaboration with international partners, focusing on the most important aspects related to ChatGPT. This approach ensured that the questionnaire addressed diverse perspectives and captured a comprehensive view of students’ perceptions of ChatGPT in the context of higher education. The preliminary version of the questionnaire was validated with students from Slovenia (see Aristovnik et al. [11]), and the final version was refined based on feedback from pilot testing, enhancing its reliability and relevance for the target population. In order to ensure and achieve a global reach, the questionnaire was later translated into six additional languages, including Italian, Spanish, Turkish, Japanese, Arabic, and Hebrew, by native speakers proficient in English.

Limitations and uncertainty · source passage

Although the study’s large and diverse global sample of students is a notable strength of our research, several limitations must be acknowledged. First, the use of convenience sampling for recruiting participants led to uneven representation across various socio-demographic subgroups, with geographical coverage being a prime example. Despite including participants from over 100 countries and/or territories, less than 1% came from low income countries. Therefore, some findings may be biased to some extent, and caution should be exercised when generalizing the results to countries and/or territories not adequately represented in the sample. Second, the study captured only early impressions and experiences of students with ChatGPT. As generative AI technologies evolve and students become more familiar with their strengths and limitations, these initial impressions may not fully align with their future opinions on ChatGPT. Third, the questionnaire relied on students’ self-reports, which can be subject to information bias. Therefore, it is plausible that some students may have either underestimated or overestimated their early perceptions of ChatGPT in various aspects. Finally, the identified socio-demographic and geographic differences in students’ early perceptions may also reflect factors beyond ChatGPT that were not covered in the questionnaire, such as variations in the digital transformation of higher education, economic development, cultural and religious backgrounds, and political circumstances.

Design the system before the pilot

Use this study to inform a learning mechanism or implementation question. It is not direct evidence that autonomous agents improve learning. Keep the proposed agent behavior separate from what the source actually tested.

Use common rubrics, blind scoring where feasible and a fresh unaided reasoning task. Document permissible assistance and avoid treating AI-detector scores as proof of misconduct.

A 90-day district pilot

Days 1–15 · Appraise. Have the curriculum lead and a teacher read the methods and limitations. Record the study design, sample, comparator, outcomes and evidence gaps. Define a local question that the study can help inform.
Days 16–30 · Design. Choose one disciplinary task. Require a first attempt, a documented AI consultation, a critique of the output and an independent explanation. Align the comparison group and scoring rubric with the same objective. Name an accountable owner and set the baseline, success threshold and budget before implementation.
Days 31–60 · Test. Start with a small voluntary cohort, preserve an accessible non-AI route and require teacher review. Use synthetic or de-identified examples; approve data handling and tool terms through district processes.
Days 61–90 · Decide. Blind-score reasoning and accuracy on a new task without AI. Distinguish model examination scores, student perceptions and actual learner gains; only the last directly tests educational benefit. Compare with baseline, document uncertainty and stop or revise if learning worsens, access is unequal or critical errors persist.

The board decision brief

Report the learning problem, evidence quality, participating population, total cost including review time, measured benefit, errors, access differences and remaining uncertainties. Make a dated recommendation to stop, revise or expand. A positive perception survey alone does not establish learning improvement.

Attribution and reuse

Dejan Ravšelj; Damijana Keržič; Nina Tomaževič; Lan Umek; Nejc Brezovar; Noorminshah A Iahad; Ali Abdulla Abdulla; Anait Akopyan; Magdalena Waleska Aldana Segura; Jehan AlHumaid; Mohamed Farouk Allam; Maria Alló; Raphael Papa Kweku Andoh; Octavian Andronic; Yarhands Dissou Arthur; Fatih Aydın; Amira Badran; Roxana Balbontín-Alvarado; Helmi Ben Saad; Andrea Bencsik; Isaac Benning; Adrian Besimi; Denilson da Silva Bezerra; Chiara Buizza; Roberto Burro; Anthony Bwalya; Cristina Cachero; Patricia Castillo-Briceno; Harold Castro; Ching Sing Chai; Constadina Charalambous; Thomas K F Chiu; Otilia Clipa; Ruggero Colombari; Luis José H Corral Escobedo; Elísio Costa; Radu George Crețulescu; Marta Crispino; Nicola Cucari; Fergus Dalton; Meva Demir Kaya; Ivo Dumić-Čule; Diena Dwidienawati; Ryan Ebardo; Daniel Lawer Egbenya; MoezAlIslam Ezzat Faris; Miroslav Fečko; Paulo Ferrinho; Adrian Florea; Chun Yuen Fong; Zoë Francis; Alberto Ghilardi; Belinka González-Fernández; Daniela Hau; Md Shamim Hossain; Theo Hug; Fany Inasius; Maryam Jaffar Ismail; Hatidža Jahić; Morrison Omokiniovo Jessa; Marika Kapanadze; Sujita Kumar Kar; Elham Talib Kateeb; Feridun Kaya; Hanaa Ouda Khadri; Masao Kikuchi; Vitaliy Mykolayovych Kobets; Katerina Metodieva Kostova; Evita Krasmane; Jesus Lau; Wai Him Crystal Law; Florin Lazăr; Lejla Lazović-Pita; Vivian Wing Yan Lee; Jingtai Li; Diego Vinicio López-Aguilar; Adrian Luca; Ruth Garcia Luciano; Juan D Machin-Mastromatteo; Marwa Madi; Alexandre Lourenço Manguele; Rubén Francisco Manrique; Thumah Mapulanga; Frederic Marimon; Galia Ilieva Marinova; Marta Mas-Machuca; Oliva Mejía-Rodríguez; Maria Meletiou-Mavrotheris; Silvia Mariela Méndez-Prado; José Manuel Meza-Cano; Evija Mirķe; Alpana Mishra; Ondrej Mital; Cristina Mollica; Daniel Ionel Morariu; Natalia Mospan; Angel Mukuka; Silvana Guadalupe Navarro Jiménez; Irena Nikaj; Maria Mihaylova Nisheva; Efi Nisiforou; Joseph Njiku; Singhanat Nomnian; Lulzime Nuredini-Mehmedi; Ernest Nyamekye; Alka Obadić; Abdelmohsen Hamed Okela; Dorit Olenik-Shemesh; Izabela Ostoj; Kevin Javier Peralta-Rizzo; Almir Peštek; Amila Pilav-Velić; Dilma Rosanda Miranda Pires; Eyal Rabin; Daniela Raccanello; Agustine Ramie; Md Mamun ur Rashid; Robert A P Reuter; Valentina Reyes; Ana Sofia Rodrigues; Paul Rodway; Silvia Ručinská; Shorena Sadzaglishvili; Ashraf Atta M S Salem; Gordana Savić; Astrid Schepman; Samia Mokhtar Shahpo; Abdelmajid Snouber; Emma Soler; Bengi Sonyel; Eliza Stefanova; Anna Stone; Artur Strzelecki; Tetsuji Tanaka; Carolina Tapia Cortes; Andrea Teira-Fachado; Henri Tilga; Jelena Titko; Maryna Tolmach; Dedi Turmudi; Laura Varela-Candamio; Ioanna Vekiri; Giada Vicentini; Erisher Woyo; Özlem Yorulmaz; Said A S Yunus; Ana-Maria Zamfir; Munyaradzi Zhou; Aleksander Aristovnik. Higher education students’ perceptions of ChatGPT: A global study of early reactions. PLOS ONE (2025-02-05). https://doi.org/10.1371/journal.pone.0315011

Companion prepared 2026-11-01. Original authorship and DOI are preserved. The abstract is an attributed source extract; the implementation plan is new AI-assisted editorial material, offered under the same CC BY 4.0 license. This guide has not undergone independent peer review.