The DOI matches its Crossref record; 10 of the 10 references it cites were found in a registry or an index; 1 of the 42 sources its claims cite were checked for support.
Guidance for generative AI in education and research
Overview
CoChat Summary
TL;DR
UNESCO’s 2023 guidance argues that generative AI in education and research should be governed by a human-centred approach that protects human agency, inclusion, equity, and cultural and linguistic diversity. It maps GenAI technologies and risks, proposes regulatory steps and policy measures, and offers co-design frameworks for ethical and pedagogically appropriate use.
Key findings
- GenAI can generate text, images, video, music and code, but it does not understand the real world or the text it produces; outputs can be inaccurate, biased or harmful and must be critically evaluated.
- Key controversies include worsening digital poverty, regulation lagging behind development, use of content without consent, unexplainable “black box” models, AI-generated content polluting the Internet, lack of real-world understanding, marginalization of minority voices, and deeper deepfakes.
- As of 2023, 137 of 194 countries had data protection legislation; 67 countries had developed or planned national AI strategies; only around 20 had clear AI ethics regulations; only China had released specific official GenAI regulation as of July 2023; only 7 countries reported AI teacher-training frameworks; 15 had government-endorsed AI curricula; and 4 reported validated AI tools for learners with disabilities.
- The guidance recommends regulatory steps for governments, ten accountabilities for GenAI providers, institutional and individual-user actions, eight policy measures for education and research, and co-design frameworks for research, teaching, 1:1 coaching, inquiry/project-based learning, and learners with special needs; it recommends a minimum age of 13 for independent conversations with GenAI platforms.
Methods
- Design: UNESCO guidance/policy document, not an empirical study; it builds on UNESCO’s 2021 Recommendation on the Ethics of AI, the 2019 Beijing Consensus on AI and Education, and the 2022 AI and education guidance for policy-makers.
- Data/sample: draws on a UNESCO 2023 mapping/review of governmental strategies for regulating GenAI, a UNESCO survey to 193 Member States on governmental use of AI in education, OECD AI Policy Observatory, Stanford University’s AI Index Report, and expert inputs; it also includes illustrative tables of GenAI tools and co-design frameworks.
- Analysis: narrative synthesis and policy recommendation, with no primary experimental data, statistical modelling, or systematic review methodology reported.
Limitations
- The paper states that GenAI impacts are just beginning to unfold, long-term effects are not fully understood, and many proposed educational uses are “potential but unproven”; it also notes little evidence that some applications, such as conversational diagnosis or AI-detection of plagiarism, are effective or safe.
- The text notes that dominant GenAI providers often do not allow rigorous independent academic review, and that regulation lags behind rapid technological development.
- Evident from the text: this is a normative guidance document rather than an empirical evaluation, so its recommendations rely on expert consensus, limited survey data, and illustrative examples rather than tested causal evidence.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 17 Not valid · 42 Not checked · 86 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Dominant GenAI providers have also been criticized for not allowing their systems to be subject to rigorous independent academic review (Dwivedi et al., 2023).”
What the sources say
- Could not verifycites Dwivedi et al., 2023 sourceExperimental — accuracy not yet measured
No passage in our copy states that dominant GenAI providers have been criticized for refusing to let their systems undergo rigorous independent academic review. The nearest material is Grover's remark (4.2.10.1, p. 34) that the algorithms are 'usually a black box behind a wall of corporate control' whose lack of transparency complicates governance, and Contribution 23 (p. 32) warning that 'those in control of language models can manipulate the answers given' — both about opacity/control generally, not about blocking academic scrutiny. Contribution 30 (p. 40) in fact stresses that ChatGPT is an 'open access system' that everyone can use, and Contribution 26 (p. 36) discusses journal/conference policies (e.g., ICML) on AI-generated text rather than provider access. Searched: independent academic review, researcher access/auditing of closed models, corporate control and transparency, big-tech gatekeeping; our copy is also flagged as truncated at the end. (LLM, conf=0.68)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
In the paper
“While in the EU, the proposed EU AI Act requires AI tools developers to disclose the copyrighted materials they used in building their systems (European Commission, 2021).”
What the sources say
- Not matchedcites European Commission, 2021
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
In the paper
“However, according to UNESCO data, only some 15 countries had developed and implemented, or were in the process of developing government-endorsed AI curricula in schools in early 2022 (UNESCO, 2022 c ).”
What the sources say
- Not matchedcites UNESCO, 2022 c
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
None of the sources cited for this claim have been checked yet.
In the paper
“Indeed, only around twenty countries had defined any clear regulations on the ethics of AI including as they relate to education, either as part of national AI strategies or otherwise.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“Only the Ministry of Education of Singapore reported building an online repository centred on the use of ChatGPT in teaching and learning.”
What the sources say
- Not matchedcites UNESCO, 2023 c
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
28 more claims
Create a free account to see all 145 claims and to verify the 2 references still open.
- Probably not a substantial claim. 98
- It may state two separate claims as one. 49
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“UNESCO's [48] global guidance treats generative artificial intelligence as a matter of governance as much as pedagogy, precisely because these systems introduce risks that cannot be addressed by classroom technique alone, including risks to privacy, intellectual integrity, and the conditions of equitable access to reliable knowledge.”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [48]
The cited source says what the claim attributes to it.
In the citing paper
“Holmes and Miao (2023) [13] and Chauncey and McKenna (2023) [14] draw attention to a significant concern in these cases: not simply the potential for dishonesty, but the risk of normalizing knowledge production that is rhetorically convincing yet lacking in intellectual substance.”
From Generative AI and Cognitive Offloading in Higher Education: Implications for Learning, Knowledge Production, and Academic Thinking (2026) · cites this paper as [13]
In the citing paper
“UNESCO's [48] guidance treats this as a governance and ethics issue as much as a pedagogical one, precisely because the epistemic risks are inseparable from questions of trust, accountability, and equity.”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [48]
In the citing paper
“The UNESCO guidance on generative AI (2023) warns against using AI in a way that can increase discrimination against students who do not speak dominant languages [28].”
From The impact of generative artificial intelligence tools on assessment equity for non-native English-speaking medical students: a systematic review (2026) · cites this paper as [28]
In the citing paper
“UNESCO guidance states that generative AI systems run a risk of favoring culturally dominant languages and values unless those systems are subject to bias audits and/or multilingual validation [28].”
From The impact of generative artificial intelligence tools on assessment equity for non-native English-speaking medical students: a systematic review (2026) · cites this paper as [28]
3 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 3.2 Steps to Regulate GenAI in education · p. 21
“The review suggests a series of six steps that governmental agencies can take to regulate generative AI and reassert public control in order to leverage its potentials across sectors, including in education.”
§ Step 7: Reflect on the long-term implications of GenAI for education and research · p. 23
“## Step 7: Reflect on the long-term implications of GenAI for education and research”
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 1
- European Union. 2016. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) . Brussels, Official Journal of the European Union. Available at: http://data.europa. eu/eli/reg/2016/679/oj (Accessed 23 June 2023.)
Cited by
The promise and challenges of generative AI in education
Giannakos, M., Azevedo, R., Brusilovsky, P., Cukurova, M., Dimitriadis, Y., Hernandez-Leo, D., Järvelä, S., Mavrikis, M., & Rienties, B. (2025) · Behaviour & Information Technology · doi:10.1080/0144929x.2024.2394886 · Cited by 543
Jin, Y., Yan, L., Echeverria, V., Gašević, D., & Martinez-Maldonado, R. (2025) · Computers and Education Artificial Intelligence · doi:10.1016/j.caeai.2024.100348 · Cited by 452
Exploring students' perspectives on generative AI-assisted academic writing
Kim, J., Yu, S., Detrick, R., & Li, N. (2025) · Education and Information Technologies · doi:10.1007/s10639-024-12878-7 · Cited by 438
Kong, S.-C., & Yang, Y. (2024) · IEEE Transactions on Learning Technologies · doi:10.1109/tlt.2024.3392830 · Cited by 298
Perkins, M., Furze, L., Roe, J., & MacVaugh, J. (2024) · Journal of University Teaching and Learning Practice · doi:10.53761/q3azde36 · Cited by 291
Metadata
- Venue
- UNESCO eBooks
- Publisher
- UNESCO
- Published
- 2023
- OpenAlex
- W4395957952
- Cited by
- 808
- References
- 0
- Citations by year
- 2026: 408 · 2025: 236 · 2024: 152 · 2023: 12
- FWCI
- 273.67
- Type
- book
- Language
- en
- Open access
- closed
- Locations
- UNESCO eBooks (publishedVersion)
- Topics
- Social Sciences › Education › Artificial Intelligence in Education
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- generative AI, artificial intelligence in education, artificial intelligence in research, responsible AI use, AI ethics
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-11-13, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-13 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Guidance for generative AI in education and research. https://app.cochat.ai/papers/10.54675/ewzm9535
- The paper
- Guidance for generative AI in education and research. (2023). UNESCO eBooks. https://doi.org/10.54675/ewzm9535
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.