CoChat

Promises and challenges of generative artificial intelligence for human learning

Open access
Yan, Lixiang, Greiff, Samuel, Teuber, Ziwen, Gašević, Dragan · 2024 · Nature Human Behaviour

The DOI matches its Crossref record; 114 of the 114 references it cites were found in a registry or an index; 38 of the 96 sources its claims cite were checked for support.

Abstract

Generative artificial intelligence (GenAI) holds the potential to transform the delivery, cultivation and evaluation of human learning. Here the authors examine the integration of GenAI as a tool for human learning, addressing its promises and challenges from a holistic viewpoint that integrates insights from learning sciences, educational technology and human-computer interaction. GenAI promises to enhance learning experiences by scaling personalized support, diversifying learning materials, enabling timely feedback and innovating assessment methods. However, it also presents critical issues such as model imperfections, ethical dilemmas and the disruption of traditional assessments. Thus, cultivating AI literacy and adaptive skills is imperative for facilitating informed engagement with GenAI technologies. Rigorous research across learning contexts is essential to evaluate GenAI's effect on human cognition, metacognition and creativity. Humanity must learn with and about GenAI, ensuring that it becomes a powerful ally in the pursuit of knowledge and innovation, rather than a crutch that undermines our intellectual abilities.

Overview

CoChat Summary

TL;DR

This Perspective argues that generative AI can transform human learning by scaling personalised support, diversifying resources, enabling timely feedback, and innovating assessment, but that these promises are counterbalanced by model imperfections, ethical dilemmas, and the disruption of conventional assessment. The authors call for AI literacy, evidence-based decision-making, and methodological rigour so that GenAI augments rather than undermines human cognition, metacognition, and creativity.

Key findings

  • GenAI promises to act as a "master teacher at scale," providing personalised, adaptive support within learners' Zone of Proximal Development; Khan Academy's GPT-4-powered "Khanmigo" is cited as a representative case that guides students through concepts rather than giving direct answers.
  • On resources, GPT-4 generated college-level biology questions well at lower Bloom's Taxonomy levels (remember, understand) but struggled at higher levels (apply, create); a preliminary study found no significant differences in learning gains or perceived experience between GenAI-generated videos with synthetic instructors and traditional recorded instructor videos.
  • On ethics, a cited systematic review found that 92% of GenAI tools used to support learning are transparent only to AI experts, not to educators, students, or other stakeholders; a divergence attack on ChatGPT reportedly caused it to output original training data containing personally identifiable information.
  • On assessment, a randomised controlled experiment found that GenAI tools helped students achieve better performance but that removing this support significantly lowered their performance — an "illusion of improved learning" that may not develop skills such as self-regulated learning.

Methods

  • Design: a Perspective paper, not an empirical study; it integrates insights from learning sciences, educational technology, and human–computer interaction to examine GenAI as a tool for human learning.
  • Structure: organised around four promises (learning support, resources, feedback, assessment), three challenges (GenAI's imperfections, ethical dilemmas, disruption of assessment), and three needs (AI literacy, evidence-based decision-making, methodological rigour), with two summary tables and two figures.
  • Evidence base: synthesis of prior literature and cited studies rather than new data collection or analysis; the authors propose a future research agenda including longitudinal and randomised controlled studies comparing GenAI tutoring with conventional rule-based tutoring.

Limitations

  • Stated: the authors acknowledge that empirical evidence on the short- and long-term impacts of GenAI on learning outcomes is currently limited, and that emerging evidence on engagement, agency, and performance is mixed (e.g., lack of learning gains after removing GenAI supports); they call for longitudinal and randomised controlled studies across academic terms and subjects.
  • Stated: much effort is required to develop valid and reliable behavioural and psychological indicators in novel GenAI-based assessment settings, and more research is needed on the optimal level of automation versus human control, expert validation, and alignment with learning objectives.
  • Evident from the text: as a Perspective, the paper presents no primary data, samples, or statistical analyses of its own; its claims rest on cited secondary studies, several of which are preprints or non-peer-reviewed sources.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41562-024-02004-5

Cites

114 of 114 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

34 Valid · 1 Not valid · 38 Not checked · 12 Own results and interpretations

53 more claims · 22 with their cited sources checked

Create a free account to see all 85 claims and to verify the 16 references still open.

See all claims — free account
  • It may state two separate claims as one. 25
  • Probably not a substantial claim. 3

Create a free account to see these claims.

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?

4 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Cited works that were retracted

Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.

2 of 2

Citations

Cites

2 retracted

Show all 114 references

Research works printed but not linked: 8

  1. Vygotsky, L. S. & Cole, M. Mind in Society: Development of Higher Psychological Processes (Harvard Univ. Press, 1978).
  2. Darvishi, A., Khosravi, H., Sadiq, S., Gašević, D. & Siemens, G. Impact of AI assistance on student agency. Comput. Educ. 210, 104967 (2024). · DOI · Paper page
  3. Chen, Y., Zhang, X. & Hu, L. A progressive prompt-based image-generative AI approach to promoting students’ achievement and perceptions in learning ancient Chinese poetry. Educ. Technol. Soc. 27, 284–305 (2024).
  4. Bada, S. O. & Olusegun, S. Constructivism learning theory: a paradigm for teaching and learning. J. Res. Method Educ. 5, 66–70 (2015).
  5. McCarthy, J. Evaluating written, audio and video feedback in higher education summative assessment tasks. Issues Educ. Res. 25, 153–169 (2015).
  6. Winograd, A. Loose-lipped large language models spill your secrets: the privacy implications of large language models. Harvard J. Law Technol. 36, 616–656 (2023).
  7. Sha, L., Li, Y., Gasevic, D. & Chen, G. Bigger data or fairer data? Augmenting BERT via active sampling for educational text classification. In Proc. 29th International Conference on Computational Linguistics 1275–1285 (International Committee on Computational Linguistics, 2022).
  8. Choi, J. H., Hickman, K. E., Monahan, A. B. & Schwarcz, D. ChatGPT goes to law school. J. Leg. Educ. 71, 387 (2021).

Cited by

See all 499

Metadata

Authors
Yan, L., Greiff, S., Teuber, Z., & Gašević, D.
Affiliations
  • Monash University (AU)
  • Goethe University Frankfurt (DE)
  • University of Luxembourg (LU)
Corresponding author
  • Greiff, Samuel
  • Gašević, Dragan
Venue
Nature Human Behaviour
Published
2024-10-22
Volume
8
Issue
10
Pages
1839-1850
ISSN
2397-3374 (ISSN-L)
OpenAlex
W4403637324
Cited by
499
References
107
Citations by year
2026: 286 · 2025: 203 · 2024: 9
FWCI
28.60
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Human Behaviour (publishedVersion)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • Open Repository and Bibliography (University of Luxembourg) (submittedVersion, other-oa)
  • HAL (submittedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • arXiv (Cornell University) (cc-by)
Funders
  • Australian Government
  • Defense Advanced Research Projects Agency (HR0011-22-2-0047)
  • Advanced Research Projects Agency
  • Australian Research Council (DP220101209, DP240100069)
Topics
  • Social Sciences › Education › Artificial Intelligence in Education
  • Physical Sciences › Computer Science › Artificial Intelligence › Intelligent Tutoring Systems and Adaptive Learning
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
generative AI, human learning, personalized learning, learning materials, feedback, assessment, AI literacy, AI ethics, model imperfections, metacognition, creativity, human-AI interaction, educational technology
MeSH terms
Artificial Intelligence, Humans, Learning

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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and 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: Promises and challenges of generative artificial intelligence for human learning. https://app.cochat.ai/papers/10.1038/s41562-024-02004-5
The paper
Yan, L., Greiff, S., Teuber, Z., & Gašević, D. (2024). Promises and challenges of generative artificial intelligence for human learning. Nature Human Behaviour. https://doi.org/10.1038/s41562-024-02004-5

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.