AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking
Open accessThe DOI matches its Crossref record; 34 of the 34 references it cites were found in a registry or an index; 5 of the 47 sources its claims cite were checked for support.
Abstract
The proliferation of artificial intelligence (AI) tools has transformed numerous aspects of daily life, yet its impact on critical thinking remains underexplored. This study investigates the relationship between AI tool usage and critical thinking skills, focusing on cognitive offloading as a mediating factor. Utilising a mixed-method approach, we conducted surveys and in-depth interviews with 666 participants across diverse age groups and educational backgrounds. Quantitative data were analysed using ANOVA and correlation analysis, while qualitative insights were obtained through thematic analysis of interview transcripts. The findings revealed a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading. Younger participants exhibited higher dependence on AI tools and lower critical thinking scores compared to older participants. Furthermore, higher educational attainment was associated with better critical thinking skills, regardless of AI usage. These results highlight the potential cognitive costs of AI tool reliance, emphasising the need for educational strategies that promote critical engagement with AI technologies. This study contributes to the growing discourse on AI’s cognitive implications, offering practical recommendations for mitigating its adverse effects on critical thinking. The findings underscore the importance of fostering critical thinking in an AI-driven world, making this…
Overview
CoChat Summary
TL;DR
In a UK mixed-methods study of 666 valid participants, greater AI tool usage was associated with lower critical thinking, with cognitive offloading partially mediating this relationship. Younger participants and those with lower educational attainment showed higher AI reliance/offloading and lower critical thinking, while higher education was linked to better critical thinking.
Key findings
- Correlation analysis reported strong associations: AI tool use ↔ cognitive offloading r = +0.72; AI tool use ↔ critical thinking r = −0.68; cognitive offloading ↔ critical thinking r = −0.75 (Table 6). Table 5 also reports smaller correlations (AI tool use–critical thinking −0.49; cognitive offloading–critical thinking −0.48).
- Mediation analysis found cognitive offloading significantly mediated the AI-use–critical-thinking relationship: total effect b = −0.42 (SE = 0.08, p < 0.001), indirect effect b = −0.25 (SE = 0.06, p < 0.001), direct effect b = −0.17 (SE = 0.05, p < 0.01).
- Multiple regression showed AI tool use negatively predicted critical thinking (β = −1.76, p < 0.001), while education level (β = 0.33, p < 0.001) and deep thinking activities (β = −0.36, p < 0.001) were also significant predictors; random forest regression explained 37% of variance (R² = 0.370) and the permutation test was significant (p = 0.0099).
- Younger participants (17–25) exhibited higher AI tool usage and cognitive offloading but lower critical thinking scores; older participants (46+) showed lower AI usage/offloading and higher critical thinking. Qualitative themes were AI dependence, cognitive engagement, and ethical concerns.
Methods
- Mixed-method design: survey of 669 UK participants (666 valid) using a 23-item questionnaire based on HCTA and Terenzini measures with a 6-point Likert scale, plus 50 semi-structured interviews.
- Quantitative analyses included descriptive statistics, ANOVA, Pearson correlations, multiple regression, random forest regression, and a permutation test; qualitative data were analysed using Braun and Clarke’s six-phase thematic analysis.
- Participants were recruited through convenience and purposive sampling promoted online in the UK, categorised into age groups 17–25, 26–45, and 46+.
Limitations
- Stated in the paper: reliance on self-reported measures and potential sample bias.
- Evident from the text: the cross-sectional design cannot establish causality; the UK-only convenience/purposive sample is demographically skewed (e.g., 48.5% doctorate, 43.7% aged 26–35) despite claims of diversity; and correlation values differ between Table 5 and Table 6.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 1 Not valid · 34 Not checked · 87 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Zhai et al. [25] found that students who heavily relied on AI dialogue systems exhibited diminished decision-making and critical analysis abilities, as these systems allowed them to offload essential cognitive tasks.”
What the sources say
- Supportedcites [25] Zhai, 2024, The effects of over-reliance on AI dialogue systems on students’ cognitive abilities: A systematic review, Smart Learn. Environ. source
The Introduction states that overdependence on AI-generated content often reduces critical and analytical thinking skills. The Discussion reports degraded decision-making abilities among some students and says replacing human decision-making with AI undermines those mental practices. These passages support the citing sentence as a summary of the review’s findings. (LLM, conf=0.88)
plainstates a cause- 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.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Our results support this approach, indicating that reducing cognitive offloading through active engagement can mitigate the negative impact of AI tools on critical thinking.”
What the sources say
- Not checked — no full text of the sourcecites Freeman et al. [45] Freeman, 2014, Active learning increases student performance in science, engineering, and mathematics, Proc. Natl. Acad. Sci. USA source
- Partly supportedcites Deslauriers et al. [46] Deslauriers, 2019, Measuring actual learning versus feeling of learning in response to being actively engaged in the classroom, Proc. Natl. Acad. Sci. USA sourceExperimental — accuracy not yet measured
The Results and Discussion section links active engagement and increased cognitive effort to deeper learning than passive instruction. It does not examine AI tools, cognitive offloading, or whether active engagement mitigates an AI-related decline in critical thinking. The citing authors’ reference to “our results” is their own conclusion, not a finding of this source. (LLM, conf=0.86)
hedgedstates a cause- 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.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Our findings are consistent with previous research indicating that excessive reliance on AI tools can negatively impact critical thinking skills.”
What the sources say
- Partly supportedcites Firth et al. [24] Firth, 2019, The “online brain”: How the Internet may be changing our cognition, World Psychiatry sourceExperimental — accuracy not yet measured
In “How does a supernormal stimulus interact with normal cognition?” (page unavailable), the paper reports that less analytical thinkers used smartphones more for online information searches and discusses possible cognitive miserliness. It also suggests reliance on online searching may impede memory retrieval. These passages concern Internet use, not AI tools, and do not show that reliance on AI harms critical-thinking skills. (LLM, conf=0.82)
- Not matchedcites Zhang et al. [43] Zhang, 2021, Automatically detecting cognitive engagement beyond behavioral indicators: A case of online professional learning community, Educ. Technol. Soc.
We searched by its title and found no matching paper.
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.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The findings revealed a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading.”
plainstates a cause- 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.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The findings revealed a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading.”
plainstates an association- 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.
71 more claims
Create a free account to see all 126 claims and every source verdict.
- Probably not a substantial claim. 25
- It may state two separate claims as one. 32
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 source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent evidence suggests that students frequently accept AI-generated content with limited verification, struggle to assess the validity of its outputs, and over-rely on these tools even in their own areas of expertise [9, 17].”
From Students' Practices and Skills in the LLM-Era: "You Can't Outsource the Struggle and Still Get the Skill" (2026) · cites this paper as [9]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent evidence suggests that students frequently accept AI-generated content with limited verification, struggle to assess the validity of its outputs, and over-rely on these tools even in their own areas of expertise [9, 17].”
From Students' Practices and Skills in the LLM-Era: "You Can't Outsource the Struggle and Still Get the Skill" (2026) · cites this paper as [9]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent evidence suggests that students frequently accept AI-generated content with limited verification, struggle to assess the validity of its outputs, and over-rely on these tools even in their own areas of expertise [9, 17].”
From Students' Practices and Skills in the LLM-Era: "You Can't Outsource the Struggle and Still Get the Skill" (2026) · cites this paper as [9]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Educators must understand accountability in knowledge production and the effects of AI on critical inquiry and metacognition (Gerlich, 2025;Jose, Cleetus, et al., 2025;Jose, Cherian, et al., 2025).”
From An integrated AI literacy framework for higher‑education teachers: a critical review and conceptual synthesis (2026) · cites this paper as Gerlich, 2025
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Several recent studies have found a negative correlation between GenAI use and critical thinking skills, confirming educators’ concern that the availability of technologies like ChatGPT may undermine development of critical thinking among college students [125, 127, 128].”
From Autonomous artificial intelligence, scientific research, and human values (2026) · cites this paper as [125]
Experimental — accuracy not yet measured
14 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 major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 4.3. Correlation Analysis
“| AI Tool Use | 1.00 | 0.89 | - 0.49 | | Cognitive Offloading | 0.89 | 1.00 | - 0.48 |”
§ 4.3. Correlation Analysis
“| AI Tool Use ↔ Cognitive Offloading | +0.72 | Strong positive correlation | | AI Tool Use ↔ Critical Thinking | - 0.68 | Strong negative correlation | | Cognitive Offloading ↔ Critical Thinking | - 0.75 | Strong negative correlation |”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
“| Age | 91.65 | 4 | 22.91 | 12.92 | <0.001 |”
Reported 12.92 · recomputed 91.64
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 4.4. Multiple Regression
“Higher education levels and deep thinking activities positively influence critical thinking, while increased AI tool usage has a detrimental effect.”
§ 4.4. Multiple Regression
“| Deep Thinking Activities | - 0.36 | 0.08 | - 4.50 | <0.001 |”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 4.2. ANOV A
“The ANOVA results revealed significant differences in critical thinking scores across different levels of AI tool usage ( p < 0.001)”
§ 4.2. ANOV A
“| Education Level | 1053.71 | 3 | 351.24 | 1401.81 | <0.001 | | Gender | 0.04 | 1 | 0.04 | 0.14 | 0.71 | | Occupation | 38.73 | 3 | 12.91 | 6.98 | <0.001 | | Age | 91.65 | 4 | 22.91 | 12.92 | <0.001 |”
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
§ 4.7.2. Critical Thinking and Educational Background
“One participant noted, 'I use AI because it simplifies everything, but I sometimes feel like I'm losing my own problem-solving skills' (P221, high school graduate).”
§ Frequencies for Gender
“| 2 (Some college) | 46 | 6.907 | | 3 (Bachelor's) | 115 | 17.267 | | 4 (Master's) | 182 | 27.327 | | 5 (Doctorate) | 323 | 48.498 | | Missing | 0 | 0.000 | | Total | 666 | 100.000 |”
1 more finding
Create a free account to see every finding of this check.
Citations
Cites
Using thematic analysis in psychology
Braun, V., & Clarke, V. (2006) · Qualitative Research in Psychology · doi:10.1191/1478088706qp063oa · Cited by 194,414
Cognitive Load During Problem Solving: Effects on Learning
Sweller, J. (1988) · Cognitive Science · doi:10.1016/0364-0213(88)90023-7 · Cited by 11,048 · cited 1×
“Cognitive load theory, developed by Sweller [19], posits that the human cognitive system has limited capacity, and reducing cognitive load can enhance learning and performance.”
Active learning increases student performance in science, engineering, and mathematics
Freeman, S., Eddy, S. L., McDonough, M., Smith, M. K., Okoroafor, N., Jordt, H., & Wenderoth, M. P. (2014) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1319030111 · Cited by 9,787 · cited 1×
“Freeman et al. [45] and Deslauriers et al. [46] advocate for active learning strategies that involve students in the learning process.”
SAGE Handbook of Mixed Methods in Social & Behavioral Research
Tashakkori, A., & Teddlie, C. (2010) · doi:10.4135/9781506335193 · Cited by 6,215
Likert scales, levels of measurement, and the “laws” of statistics
Norman, G. (2010) · Advances in Health Sciences Education · doi:10.1007/s10459-010-9222-y · Cited by 4,850 · cited 1×
“A 6-step Likert scale (1 = Never/Strongly Disagree to 6 = Always/Strongly Agree) was used for most items, providing ordinal data suitable for parametric statistical analysis [34].”
Lazer, D. M. J., Baum, M. A., Benkler, Y., Berinsky, A. J., Greenhill, K. M., Menczer, F., Metzger, M. J., Nyhan, B., Pennycook, G., Rothschild, D., et al. (2018) · Science · doi:10.1126/science.aao2998 · Cited by 4,160 · cited 1×
“For example, AI-driven news aggregators can personalise news feeds based on the credibility of sources, helping users access more reliable information [22].”
The Black Box Society: The Secret Algorithms That Control Money and Information
Pasquale, F. (2015) · doi:10.4159/harvard.9780674736061 · Cited by 3,796
FILTER BUBBLE: Wie wir im Internet entmündigt werden
Pariser, E. (2012) · doi:10.3139/9783446431164 · Cited by 1,849
Google Effects on Memory: Cognitive Consequences of Having Information at Our Fingertips
Sparrow, B., Liu, J., & Wegner, D. M. (2011) · Science · doi:10.1126/science.1207745 · Cited by 1,762
Intelligence and educational achievement
Deary, I. J., Strand, S., Smith, P., & Fernandes, C. (2007) · Intelligence · doi:10.1016/j.intell.2006.02.001 · Cited by 1,723 · cited 1×
“…‘education level’ and ‘frequency of deep thinking activities’ are also included, as they are recognised predictors of cognitive engagement and critical thinking, as highlighted in previous research linking educational attainment and intellectual engagement with critical reasoning skills [41,42].”
Research works printed but not linked: 8
- Paul, R., and Elder, L. (2008). The Miniature Guide to Critical Thinking Concepts and Tools, Foundation for Critical Thinking Press.
- Fisher, A. (2011). Critical Thinking: An Introduction, Cambridge University Press.
- Jonassen, D.H. (2000). Learning to Solve Problems: A Handbook for Designing Problem-Solving Learning Environments, Routledge.
- Bennett, 2021, AI and analytics: Implications for the future of work, J. Manag. Inf. Decis. Sci., 24, 1
- Clark, R.C., Nguyen, F., and Sweller, J. (2016). Efficiency in Learning: Evidence-Based Guidelines to Manage Cognitive Load, Wiley.
- Creswell, J.W., and Plano Clark, V.L. (2011). Designing and Conducting Mixed Methods Research, Sage Publications.
- Field, A. (2013). Discovering Statistics Using IBM SPSS Statistics, Sage Publications.
- Tabachnick, B.G., and Fidell, L.S. (2013). Using Multivariate Statistics, Pearson. [6th ed.].
Cited by
Items 401–500 of 2,539
From Search to Synthesis: Generative AI, Internet Use, and the Transhuman Epistemic Shift
Lockhart, E. N. S. (2025) · Journal of excellence in social sciences. · doi:10.69565/jess.v4i1.418 · Cited by 5
Choi, H., Phung, T., Wu, M., Singla, A. K., & Brooks, C. A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2512.04630 · Cited by 5
“Previous research has associated greater confidence in AI performance and more frequent AI use to increased cognitive offloading, reduced self-reported cognitive effort, and lower metacognitive engagement (Gerlich, 2025; Lee et al., 2025; Zhan and Yan, 2025).”
Training for Obsolescence? The AI-Driven Education TrapPreprint
Peterson, A. J. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2508.19625 · Cited by 5
“This aligns with findings that frequent AI tool usage can be negatively correlated with critical thinking abilities, mediated by this increase in cognitive offloading (Gerlich, 2025).”
The hidden functions of sycophancy in AI systems: steering, consistency, and cognitive dependency
Jacobowitz, S. (2026) · AI & SOCIETY · doi:10.1007/s00146-026-02993-z · Cited by 4
“Gerlich (2025) attributes these negative correlations between frequent AI tool usage and critical thinking abilities to what he terms “cognitive offloading.””
Thinking Smarter, not Harder? Google NotebookLM's Misalignment Problem in Education
Albrecht-Crane, C. (2025) · Proceedings of the 43rd ACM International Conference on Design of Communication · doi:10.1145/3711670.3764628 · Cited by 4
“…negative relationship between the frequency of students’ use of AI tools and their critical thinking abilities, a relationship “mediated by increased cognitive offloading,” particularly pronounced in younger, less-educated students (17-25 years), who reported substantial dependence on AI tools [4].”
Retnam, S., & Sadasiv, M. S. (2025) · Medical Science Educator · doi:10.1007/s40670-025-02570-x · Cited by 4
Pereza metacognitiva y descarga cognitiva en la era de la IA generativa: Riesgos y uso responsable
Mendiola, M. S. (2025) · Investigación en Educación Médica · doi:10.22201/fm.20075057e.2025.56.25743 · Cited by 4
Do neural language models critically reason about moral issues like humans?
Suriano, R., Acciai, A., Fabio, R. A., Guerrisi, L., & Plebe, A. (2026) · Current Psychology · doi:10.1007/s12144-025-08798-5 · Cited by 4
“In this area, to which some of these authors have also contributed, the results are still varied, ranging from negative effects (Gerlich, 2025; Yusuf et al., 2024 ), to decidedly positive effects (Essien et al., 2024; Zhou et al., 2024; Chen et al., 2025; Suriano et al., 2025; Fabio et al., 2025c;…”
Hansen, M., & Carrigan, M. (2025) · British Journal of Sociology of Education · doi:10.1080/01425692.2025.2519488 · Cited by 4
“Much of this has to do with reducing the friction associated with finding and structuring information, even down to mobilising the creative energy needed for a brainstorming session, or thinking deeply about a problem (Gerlich 2025).”
Pamungkas, R., Suwono, H., & Ibrohim, I. (2025) · Salud Ciencia y Tecnología · doi:10.56294/saludcyt20251976 · Cited by 4
Wu, Y. (2025) · Al lnnovations and Applications · doi:10.63944/sj51.aia · Cited by 4
“Furthermore, in language learning, automated translation and writing assistance tools reduce opportunities for learners to retrieve vocabulary, construct syntax, and practice expression, thereby limiting the development of their linguistic competencies [8] .”
LLM-“Friends” are Hostile Scaffolds in the Age of Loneliness
Sahebi, S., & Parvizi-Wayne, D. (2026) · Minds and Machines · doi:10.1007/s11023-026-09765-0 · Cited by 4
“This is supported by work from Gerlich (2025 ), which also found that heavy usage of AI tools correlated with a decline in critical thinking abilities, mediated by increased cognitive offloading onto those tools.”
Zhang, Q., Liu, L., Li, J., Xing, B., Liu, Z. H., Liu, J., Li, X., Wang, J., Wang, J., & Wang, J. (2026) · BMC Medical Education · doi:10.1186/s12909-026-09358-6 · Cited by 4
“This finding resonates with the automation bias literature, which documents how frequent use of automated systems can increase trust in their outputs while reducing engagement in independent verification [11, 14, 37, 38].”
Perdomo, K. S., & Hernández, L. G. J. (2026) · RELIEVE - Revista Electrónica de Investigación y Evaluación Educativa · doi:10.30827/relieve.v32i1.36950 · Cited by 4
Silva, E. (2025) · The Conversation · doi:10.64628/aai.ukwpcxkfr · Cited by 4
Sadidi, F., & Prestel, T. (2025) · Discover Education · doi:10.1007/s44217-025-01071-8 · Cited by 5
“Teachers are expected to critically evaluate media, considering its technological, social, and application-related aspects.”
Meylani, R. (2025) · Buca eğitim fakültesi dergisi (Online)/Buca Eğitim Fakültesi dergisi · doi:10.53444/deubefd.1527781 · Cited by 4
How does contextual fidelity impact how we think, talk, and act in AI-assisted engineering design?
Vyas, S., Cheruku, S., & Krishnamurthy, V. R. (2026) · Advanced Engineering Informatics · doi:10.1016/j.aei.2026.104456 · Cited by 4
Against Techno-Feudalism in Academia: Reclaiming Scholarly Autonomy in the Age of AI
Yang, J., Jincheng, G., & Gong, L. (2026) · IEEE Transactions on Learning Technologies · doi:10.1109/tlt.2026.3673243 · Cited by 4
“This often leads to behavioral inertia and psychological trust, which may, in turn, generate adverse effects, eroding the development and sustainabil-ityofcorecognitiveskills[16].”
Tao, S., Zhang, H., & Ding, L. (2026) · Acta Psychologica · doi:10.1016/j.actpsy.2026.106622 · Cited by 4
Using artificial intelligence as a mindtool to foster critical thinking in biology education
Mawn, M. V., & Washington, J. M. (2026) · Journal of Microbiology and Biology Education · doi:10.1128/jmbe.00048-26 · Cited by 4
Ma, L., Liu, T., Zhao, C., & Niu, T. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1844997 · Cited by 4
“Conversely, those with insufficient AIL are more vulnerable to information misjudgment, technological over-reliance, and cognitive fixation; they struggle to benefit from AI and may even fall into cognitive traps through blind trust, harming their IB (Chun et al., 2025; Gerlich, 2025).”
InfluentialZhang, H., Yang, Y., Zhang, Y., Qiu, B., & Yang, J. (2026) · BMC Psychology · doi:10.1186/s40359-026-05011-5 · Cited by 4
“In addition, passive patterns of AI use—such as uncritical copy-pasting of generated content—are associated with surface-level learning approaches and limited understanding of complex concepts [31, 32].”
Oleas, D., Rodas, J. A., & Alarcon, D. (2026) · Frontiers in Digital Health · doi:10.3389/fdgth.2026.1882677 · Cited by 4
“Within the context of AI, the central concern should lie in the progressive externalization of critical functions (13).”
A scoping review on the mental health harms of LLM-based chatbots
Diel, A., Torous, J., Cuijpers, P., Kleesiek, J., Nensa, F., Weber, N., Faust, F., Lalgi, T. J., Mellis, F. S., Teufel, M., et al. (2026) · npj Digital Medicine · doi:10.1038/s41746-026-03054-x · Cited by 4
“While some studies fi nd lower self-reported critical thinking related to LLM-CB use 86,87 or reduced critical thinking following functional LLM-CB use 88 , other studies found no effects 89 .”
You Chen, S. W. (2025) · International Journal of New Developments in Education · doi:10.25236/ijnde.2025.070509 · Cited by 4
From Use to Oversight: How Mental Models Influence User Behavior and Output in AI Writing Assistants
Rismani, S., Blodgett, S. L., Liao, Q. V., Olteanu, A., & Moon, A. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791670 · Cited by 4
Accounting Research in the Age of AIPreprint
Keloharju, M., & Keloharju, R. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5345050 · Cited by 4
Levelling Up the Literature: How Games Can Inspire Academic Reporting
Seaborn, K., & Williams, A. (2025) · Annual Symposium on Computer-Human Interaction in Play · doi:10.1145/3744736.3749331 · Cited by 4
Zhao, K., Li, L., Chen, W., & Liu, H. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2701958 · Cited by 4
Agentic Loafing: An AI Decision Delegation Risk
Issa, H., Petani, F. J., & Glavas, D. (2026) · Risk Analysis · doi:10.1111/risa.70306 · Cited by 5
“…is a risk to group productivity (Freeman and Greenacre 2011 ), and algorithmic loafing a risk to critical thinking (through cognitive-offloading) (Gerlich 2025 ), then this research argues that “ agentic loafing ” constitutes a fundamental risk to decision sovereignty, especially considering the…”
Misinformation as strategy: Epistemic consequences and the undermining of shared truth
Ecker, U. K. H., Spearing, E. R., DiResta, R., & Lewandowsky, S. (2026) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2026.06.009 · Cited by 4
Pachava, V., Lasekan, O. A., Pena, M. T. G., Swaroop, K. R., & Guniganti, S. (2026) · International Research Journal of Multidisciplinary Scope · doi:10.47857/irjms.2026.v07i02.09619 · Cited by 4
“Over-reliance, by contrast, occurs when students depend on AI-generated outputs as primary learning agents, leading to cognitive offloading and surface-level engagement (20).”
Generative AI Can Improve Performance and Engagement without Harming LearningPreprint
Kim, D., Mitrofanov, D., Wen, Q., & Xu, T. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5929576 · Cited by 4
James, T., Pike, J., Andrews, G., Valentine, J., Brickhill, M., Sharp, J., & Ulpen, T. (2025) · Journal of University Teaching and Learning Practice · doi:10.53761/40w8jg17 · Cited by 4
Sallam, M., Beaini, C., Mijwil, M. M., & Sallam, M. (2025) · Journal of Studies in Medical Sciences · doi:10.20849/jsms.v1i1.1515 · Cited by 4
“A generation raised on AI-enhanced platforms may excel in identifying what the machine missed—but at the cost of their own diagnostic development (Gerlich, 2025).”
Azman, M., Mohadih, N. B. M., Azhar, N. B. B. K., Farizan, M. A. B. A., Radzi, N. A. B. M., & Manap, M. R. (2025) · International Journal of Research and Innovation in Social Science · doi:10.47772/ijriss.2025.924ileiid0089 · Cited by 4
Generative artificial intelligence: between enhancement and cognitive offloading
Espeso-García, A. (2025) · Cultura Ciencia y Deporte · doi:10.12800/ccd.v20i66.2698 · Cited by 4
Navigating the AI Turn: Framework for Responsible AI Integration in Language Education
Pham, V. P. H., & Renandya, W. A. (2026) · International Journal of TESOL & Education · doi:10.54855/ijte.26626 · Cited by 4
“On the other hand, when AI is used to do the cognitive work for the learner, rather than support the learner, it circumvents the Zone of Proximal Development (ZPD) and creates cognitive overload: measurable output without productive struggle (Gerlich, 2025).”
Zhi, J., Long, H. J., So, R. J., & Lee, M. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791727 · Cited by 4
“Some see opportunities for enhanced efficiency in consuming and creating information [40], while others harbor concerns about cognitive outsourcing that bypasses human thought processes [20, 32, 37, 43].”
Intellectual Stewardship: Re-adapting Human Minds for Creative Knowledge Work in the Age of AIPreprint
Zhang, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.18117 · Cited by 4
“…and task completion, overreliance on AI systems may reduce student metacognitive effort, weaken their critical thinking and ownership over ideas, and foster dominant thinking patterns at the expense of idea diversity (Choi et al., 2025; Fan et al., 2025; Gerlich, 2025; Habib et al., 2024).”
Serious Games: Human-AI Interaction, Evolution, and Co-evolution
Doreswamy, N., & Horstmanshof, L. (2026) · Cureus · doi:10.7759/cureus.114096 · Cited by 4
“The burning question is whether societies will take these attributes forward or trade them for the convenience that AI offers [35].”
LLM-based In-situ Thought Exchanges for Critical Paper Reading
Fang, X., Xu, A., Yang, C.-L., Lin, Y.-F., Malacria, S., & Yatani, K. (2026) · ACM International Conference on Intelligent User Interfaces (IUI) · doi:10.1145/3742413.3789069 · Cited by 4
ChatLearn: Leveraging Non-Native Speaker Communication Challenges as Language Learning Opportunities
Qin, P., Tan, Y., Chen, J., Boonprakong, N., Zhu, Z., Yamashita, N., & Lee, Y.-C. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791839 · Cited by 4
“However, in prioritizing solving communication problems, many AIMC systems overly offload cognitive engagement [24, 68, 74], leading to weakened cognitive processing [15] and critical thinking [73] about the language challenges that NNSs encounter.”
Silitonga, L. C., Tiatri, S., Beng, J. T., Dinatha, V. O. D., Nurkholiza, R., Salsabila, T. M., Mahmud, T. N., Bunarwan, E. A., & Azzahra, C. S. (2026) · JURNAL PENDIDIKAN MIPA · doi:10.37630/jpm.v16i3.4325 · Cited by 4
Popalia, J. N., Al-Habsy, M., Akbar, M., & Affandi, A. (2026) · Artificial Intelligence in Lifelong and Life-Course Education · doi:10.66053/aillce.v1i1.3 · Cited by 4
“The selection of students was done using a purposive sampling method, which means choosing people based on certain conditions that match the study's goals (Gerlich, 2025).”
La inteligencia artificial generativa y la formación del juicio crítico en la adolescencia
Román, M. G. M., Ibarra, B. S. N., Jordán, E. R. P., Chiluisa, E. A. P., Ibarra, R. E. N., & Pérez, J. C. M. (2026) · Revista Veritas de Difusão Cientifica. · doi:10.61616/rvdc.v7i2.1638 · Cited by 4
Mendoza, K. K. R., & Cetzal, R. S. P. (2026) · Revista Educación Superior y Sociedad (ESS) · doi:10.54674/ess.v37i2.1121 · Cited by 4
Cong‐Lem, N. (2025) · TEFLIN Journal - A publication on the teaching and learning of English · doi:10.15639/teflinjournal.v36i2/279-302 · Cited by 4
“One major concern is cognitive offloading, where students could potentially become over-reliant on AI-generated responses instead of engaging in independent critical thinking (Gerlich, 2025).”
Psychological harm induced by generative AI and its legal remedies
Huang, X., Nor, M. A. N. I. M., Yusoff, Z. M., & Liang, Z. (2025) · Environment and Social Psychology · doi:10.59429/esp.v10i3.3554 · Cited by 4
Developing leader competence in situation through Artificial Intelligence and coaching
Malafronte, O. (2025) · theses.fr (ABES) · doi:10.70675/7f9b9325z0c7cz48e7zbbccz9ca3dd3c5f75 · Cited by 4
Toth, A., Roch, N., Zufferey, N., & Zimmermann, V. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-98462-4_9 · Cited by 3
When AI Thinks for Us: Unveiling the Cognitive and Social Toll of ChatGPT Over‐Reliance
Farhat, Z. (2026) · Applied Cognitive Psychology · doi:10.1002/acp.70205 · Cited by 5
Working with Large Language Models: impacts on performance, efficiency and perceived responsibility
Kares, F., Hannig, L., & Langer, M. (2026) · Behaviour and Information Technology · doi:10.1080/0144929x.2026.2666279 · Cited by 3
Mzwri, K., & Turcsányi-Szabó, M. (2025) · Research Square · doi:10.21203/rs.3.rs-6256444/v1 · Cited by 4
The mirror metaphor and AI: reflections and realities on education from an emergence perspective
Ansaldo, S. (2025) · Globalisation Societies and Education · doi:10.1080/14767724.2025.2543324 · Cited by 3
Vasylieva, M. (2025) · Contemporary Buddhism · doi:10.1080/14639947.2025.2562765 · Cited by 3
Artificial intelligence and cognitive inequality
Puri, I., & Veldkamp, L. (2025) · Journal of Monetary Economics · doi:10.1016/j.jmoneco.2025.103884 · Cited by 3
Artificial Intelligence in Schools: Opportunities, Risks, and Governance for Inclusive Learning
Zoelfakar, N., & Ibrahim, A. S. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-2600-0388-6.ch002 · Cited by 3
The Formula of Humanity and AI Use
Sticker, M. (2026) · Philosophy & Technology · doi:10.1007/s13347-026-01100-w · Cited by 3
“Moreover, there already exist a number of other arguments to the effect that writing is a form of thinking or reasoning and our capacity for reasoning may therefore be undermined when writing is delegated to an AI (Mintz, 2021; Gerlich, 2025 ).”
Castro-López, A., Abed, M. E., Cervero, A., & Iglesias, V. (2026) · Education + Training · doi:10.1108/et-10-2025-0812 · Cited by 3
21st Century Skills and Workforce Development: A Paradigm Shift
Cooper, M. M. (2026) · Journal of Chemical Education · doi:10.1021/acs.jchemed.6c00234 · Cited by 3
Positive psychology 4.0: a science of consequence for a world that can no longer wait
van Zyl, L. E. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1726065 · Cited by 3
“Gerlich (2025) found that cognitive offloading mediates a negative relationship between AI tool usage and critical thinking, particularly among younger populations.”
Ouyang, Y., Gao, S., Wang, R., Zhu, H., Shao, Y., Gu, X., & Li, Q. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790530 · Cited by 4
“A common concern with tools designed to encourage deeper thinking is that they may increase users’ operational and cognitive load [48].”
Majumder, M. A. A., & Haque, M. (2025) · Advances in Human Biology · doi:10.4103/aihb.aihb_84_25 · Cited by 3
Challenges in Ensuring Originality and Authenticity in AI-Generated Works
Kasih, E. N. E. W., Indriyanto, K., & Wardani, L. D. P. (2025) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-6330-1.ch002 · Cited by 3
Sense-Making with an AI-Enhanced Coaching Tool: A Think-Aloud Study
Sawaya, S., & D’Mello, S. K. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-98420-4_4 · Cited by 3
Xue, Z., & Fang, L. (2025) · Management Teaching Review · doi:10.1177/23792981251389862 · Cited by 3
“However, educators remain concerned about GenAI’s effects on students’ critical thinking and ethical awareness (Gerlich, 2025; Jose & Jose, 2024).”
Is ChatGPT detrimental to innovation? A field experiment among university students
Hassan, M., Amin, E., Mansour, S., & Kelani, Z. (2026) · Seikei ronsō/Review of economics and political science · doi:10.1108/reps-08-2024-0045 · Cited by 3
How Critical Thinking Shapes Student Engagement with AI Tools to Foster Self-Directed Learning
Singhachotsukpat, R., Thiangtham, K., Wilepana, C., Yaemsawat, N., Sisaeng, P., Songkram, N., & Khlaisang, J. (2025) · International Conference on Educational Communications and Technology (ICTAECT) · doi:10.1109/ictaect67351.2025.11391283 · Cited by 3
“The paradigm shift from the emergence of AI also presses a concern in contemporary education regarding the risk of students becoming overly reliant on AI, thereby reducing their engagement to passive use of these technologies [10, 11].”
Social offloading: When joint action leads to cognitive facilitation
Tufft, M. R. A., & Richardson, D. C. (2026) · Cognition · doi:10.1016/j.cognition.2026.106502 · Cited by 3
Alé-Silva, J., & Rodríguez-Donoso, M. (2026) · Discover Education · doi:10.1007/s44217-026-01131-7 · Cited by 3
The impact of AI tools: ChatGPT, Gamma and Autopilot on performance in academic assignments
Mohareb, T. W. A., & Al-Qayyem, R. (2026) · Learning Futures and Emerging Technologies · doi:10.1108/lfet-04-2025-0040 · Cited by 3
Ghimire, A., & Qiu, Y. (2026) · Nurse Education Today · doi:10.1016/j.nedt.2026.107111 · Cited by 3
AI ‐Enabled Consumer Services and Subjective Well‐Being
Pak, T. (2026) · International Journal of Consumer Studies · doi:10.1111/ijcs.70212 · Cited by 3
Wegerif, R. (2026) · Journal for the Study of Education and Development Infancia y Aprendizaje · doi:10.1177/02103702261439739 · Cited by 3
“The first warns that AI use inhibits thinking, producing “metacognitive laziness” as users offload cognitive effort (Gerlich, 2025; Lee et al., 2025).”
InfluentialBig tech, the state, or you? Who is responsible for generative AI addiction?
Stahl, B. C., Ali, R., Hitcham, L., Liebherr, M., Murray, R., Vallejos, E. P., & Yankouskaya, A. (2026) · Behaviour and Information Technology · doi:10.1080/0144929x.2026.2663509 · Cited by 3
AI and the machinization of humans: an escape from the burdens of existence
Beerends, S., Gertz, N., & Aydin, C. (2026) · AI and Ethics · doi:10.1007/s43681-026-01126-y · Cited by 3
Accuracy, reliability, and regulatory concerns in the use of AI tools in education and research
AllahRakha, N. (2026) · Social Sciences & Humanities Open · doi:10.1016/j.ssaho.2026.103102 · Cited by 3
M, S. P., David Sam Selva Jeyam, P., & Hiranmahi Gandhi, K. (2026) · Cureus · doi:10.7759/cureus.112457 · Cited by 3
“Our finding that AI dependence is negatively correlated with all domains of critical thinking, especially the Maturity and Analyticity and Flexibility domains, supports concerns regarding the impact of Generative AI on cognitive autonomy and the phenomenon of "cognitive offloading" [3,10,11].”
Generative AI in English Sixth Form Education: Student Use, Perceptions, and Literacy GapsPreprint
Ennion, M. (2025) · Research Square · doi:10.21203/rs.3.rs-7122270/v1 · Cited by 3
Researchers’ Perspectives on AI in Higher Education: A Focus Group Study
Angelis, M. C. D. (2025) · Springer eBooks · doi:10.1007/978-3-031-99754-9_7 · Cited by 3
How AI Chatbot Response Style Affects Cognitive Load and Performance in Educational Tasks
Srinivasaprasad, A., Sepanloo, K., Jahromi, S. N., Yu, D., & Duffy, V. G. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-13012-9_14 · Cited by 3
Pompei, A., Battistella, C., & Nonino, F. (2026) · International Journal of Managing Projects in Business · doi:10.1108/ijmpb-10-2025-0451 · Cited by 3
Participatory, not Punitive: Student-Driven AI Policy Recommendations in a Design Classroom
Seki, K., Vijay, M., & Kotturi, Y. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790691 · Cited by 3
Gélvez-García, L. E., & Brand, V. A. G. (2026) · Academia y Virtualidad · doi:10.18359/ravi.7925 · Cited by 3
ChatGPT Overreliance and the Erosion of L2 Writing Agency Among ESL Undergraduates in Gujarat
Makwana, V. (2026) · Journal of English Language Teaching · doi:10.66121/hdmag315 · Cited by 3
“Recent studies (Gerlich, 2025; Fan et al., 2025)”
From algorithms to tutors: tracing the evolution of generative AI in education
Aggarwal, D. (2026) · SN Social Sciences · doi:10.1007/s43545-026-01515-4 · Cited by 3
“Gerlich (2025) and others have warned of “cognitive offloading,” where students delegate critical thinking tasks to the AI, potentially eroding foundational skills.”
Lu, J., Li, W., Mo, Y., & Ge, M. (2026) · Behavioral Sciences · doi:10.3390/bs16060902 · Cited by 3
Student AI resistance: status anxiety and trust in Kazakhstani higher education
Cespedes, A. A., & Courtney, M. G. R. (2026) · Asian Education and Development Studies · doi:10.1108/aeds-11-2025-0604 · Cited by 3
Alazemi, A. M., Alenezi, A., & Alsouyan, A. (2026) · Education Sciences · doi:10.3390/educsci16071086 · Cited by 3
Li, X., Xian, X., & Du, P. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1903784 · Cited by 3
“Such use can improve short-term efficiency while weakening the student’s relationship to the product, a concern also raised in recent debates on GenAI, assessment, and academic integrity (Bittle and El-Gayar, 2025; Gerlich, 2025; Lim et al., 2023).”
An empirical study of ChatGPT use in engineering education: Prompting and performance
Domínguez, L. G. I., Robles-Gómez, A., & Pastor-Vargas, R. (2026) · The Internet and Higher Education · doi:10.1016/j.iheduc.2026.101105 · Cited by 3
Cho, M., & Merkebu, J. (2026) · New Directions for Adult and Continuing Education · doi:10.1002/ace.70031 · Cited by 3
AI as Scaffold or Shortcut? Metacognitive Regulation in Graduate Students’ Educational App Design
Almajdoub, Z., Fidalgo, P., Hashem, R., & Alramamneh, Y. (2026) · Education Sciences · doi:10.3390/educsci16081198 · Cited by 3
Deguchi, A., Tsuji, H., & Kitazawa, T. (2026) · Research in Science Education · doi:10.1007/s11165-026-10370-5 · Cited by 3
Exploring Human‐LLM Collaboration for Creativity Through the Lens of Concepts
Bai, H., Cheng, P., Duan, J., Peng, K., & Song, T. (2026) · The Journal of Creative Behavior · doi:10.1002/jocb.70132 · Cited by 3
Jenkins, D. M., Wildermuth, C. D. M. E. S., Bottomley, K., Tabata, M., & Sherlock, J. J. (2026) · New Directions for Student Leadership · doi:10.1002/yd.70069 · Cited by 3
Liu, C., Guo, J., Gao, R., Chen, X., Zhu, Q., & Chang, P. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1910545 · Cited by 3
“.2 Where this moves the debate Concerns about AI and thinking have been posed largely in terms of quantity: how much people offload (Gerlich, 2025; Zhai et al., 2024), and in the classical experimental tradition, whether and when they offload (Risko et al., 2014; Weis and Wiese, 2019; Grinschgl et al.,…”
Metadata
- Authors
- Gerlich, M.
- Affiliations
- SBS Swiss Business School (CH)
- Corresponding author
- Gerlich, Michael
- ORCID
- Gerlich, Michael
- Venue
- Societies
- Publisher
- MDPI AG
- Published
- 2025-01-03
- Volume
- 15
- Issue
- 1
- Pages
- 6
- ISSN
- 2075-4698 (ISSN-L)
- OpenAlex
- W4406026056
- Semantic Scholar
- cce6e863d5408244284d97f5a13e8c9ab103ad01
- Cited by
- 1,248
- References
- 29
- Citations by year
- 2026: 930 · 2025: 318
- FWCI
- 3829.59
- Type
- journal article
- Language
- en
- Update notices
- correction (2025-01-03)
- Open access
- gold
- doi.org
- Licence
- Locations
- Societies (publishedVersion, cc-by)
- SSRN Electronic Journal (acceptedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Social Sciences › Education › Education and Critical Thinking Development
- Social Sciences › Education › Artificial Intelligence in Education
- Social Sciences › Psychology › Developmental and Educational Psychology › Educational Strategies and Epistemologies
- Keywords
- AI tools, cognitive offloading, critical thinking, AI dependency, age differences, educational attainment, AI literacy education, mediating effect
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 2026-09-28, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
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.