From Offloading to Engagement: An Experimental Study on Structured Prompting and Critical Reasoning with Generative AI
Open accessThe DOI matches its Crossref record; 14 of the 14 references it cites were found in a registry or an index; 8 of the 46 sources its claims cite were checked for support.
Abstract
The rapid adoption of generative AI raises questions not only about its transformative potential but also about its cognitive and societal risks. This study contributes to the debate by presenting cross-country experimental data (n = 150; Germany, Switzerland, United Kingdom) on how individuals engage with generative AI under different conditions: human-only, human + AI (unguided), human + AI (guided with structured prompting), and AI-only benchmarks. Across 450 evaluated responses, critical reasoning was assessed via expert rubric ratings, while perceived reflective engagement was captured through self-report indices. Results show that unguided AI use fosters cognitive offloading without improving reasoning quality, whereas structured prompting significantly reduces offloading and enhances both critical reasoning and reflective engagement. Mediation and latent class analyses reveal that guided AI use supports deeper human involvement and mitigates demographic disparities in performance. Beyond theoretical contributions, this study offers practical implications for business and society. As organisations integrate AI into workflows, unstructured use risks undermining workforce decision making and critical engagement. Structured prompting, by contrast, provides a scalable and low-cost governance tool that fosters responsible adoption, supports equitable access to technological benefits, and aligns with societal calls for human-centric AI. These findings highlight the dual nature of generative AI as an amplifier of human capability when scaffolded and an impediment to critical thinking when adopted passively.
Overview
CoChat Summary
TL;DR
In a cross-country experiment (n = 150; Germany, Switzerland, United Kingdom), structured prompting for ChatGPT use improved critical reasoning and reflective engagement, while unguided AI use fostered cognitive offloading without improving reasoning quality. Guided AI use significantly outperformed all other conditions, and mediation/latent class analyses linked lower offloading and higher reflective engagement to better performance.
Key findings
- One-way ANOVA showed significant differences in total argument quality across the four conditions, F(3, 447) = 111.75, p < 0.0001, η² = 0.43. Tukey comparisons showed guided Human + AI significantly outperformed all other groups; guided vs. Human Alone mean difference = 7.80, p < 0.0001; unguided Human + AI vs. Human Alone = 2.56, p < 0.0001; AI Alone vs. Human Alone = 4.98, p = 0.5508 (not significant). Unguided Human + AI did not significantly outperform AI Alone.
- Cognitive offloading was negatively correlated with perceived reflective engagement (r = −0.6602, p < 0.001), education (r = −0.6165, p < 0.001), and performance (r = −0.3166, p < 0.001). Perceived reflective engagement was positively correlated with performance (r = 0.3937, p < 0.001). Offloading and perceived difficulty were not significantly correlated (r = 0.0078, p = 0.87).
- Mediation analysis found guided prompting increased perceived reflective engagement (β = 0.84, p < 0.001), which improved performance (β = 1.12, p < 0.001), with partial mediation. Sequential mediation found education predicted lower offloading (β = −0.73, p < 0.001), lower offloading predicted higher reflective engagement (β = −0.61, p < 0.001), and reflective engagement predicted performance (β = 1.12, p < 0.001).
- Latent class analysis identified two classes: Class 0 “AI-dependent thinkers” (n = 77; offloading = 4.47, reflective engagement = 2.87) and Class 1 “reflective thinkers” (n = 73; offloading = 3.83, reflective engagement = 4.49). Education had a significant multivariate effect on post-task cognition (Wilks’ Lambda = 0.1626, F = 144.34, p < 0.0001), while country did not (p = 0.076).
Methods
- Controlled multi-condition experiment with n = 150 participants (50 each in Germany, Switzerland, and the United Kingdom), recruited from academic institutions, professional workshops, and AI-focused conferences. Four conditions: human-only, AI-only, human + AI unguided, and human + AI guided. The three human-involved conditions were within-subjects and counterbalanced; AI-only was a fixed ChatGPT benchmark. The task was “What are the advantages and disadvantages of democracy?”
- Measures included expert rubric ratings of critical thinking across five dimensions (6-point scale, total 6–30; Krippendorff’s α ≥ 0.70), a post-task questionnaire on cognitive offloading, perceived reflective engagement, and perceived difficulty (6-point Likert), and 30 semi-structured interviews (10 per country).
- Analysis used ANOVA/Tukey, MANOVA, correlations, mediation and sequential mediation, latent class analysis, multiple regression, and thematic analysis, conducted in R 4.3.1 and Python 3.11, with bootstrapping (5000 samples) for robustness.
Limitations
- Stated: participants were not randomly sampled from the general population; recruitment through universities, workshops, and AI conferences may introduce self-selection bias toward more technologically engaged or reflective individuals, limiting generalisability. The accessible democracy prompt may not replicate high-stakes, domain-specific tasks. Scoring involved subjective judgement despite acceptable inter-rater reliability, and self-report measures of offloading and reflective engagement may have limited introspective access and were not a pre-validated psychometric scale. Demographic subgroup analyses were exploratory due to reduced cell sizes, and the study focused on short-term effects only.
- Stated: future research is needed across domains such as healthcare, finance, or law, and longitudinally to assess cumulative skill degradation or cognitive resilience.
- Evident from the text: the AI-only condition was a fixed benchmark generated separately rather than a randomized participant condition, limiting direct causal comparison. Some reported statistics appear inconsistent—e.g., the text states perceived reflective engagement had β = 1.12, p < 0.0001 in the regression, while Table 9 lists 0.3163, p = 0.6679; the text also says age, offloading, and perceived difficulty did not remain significant, while Table 9 and Section 3.4.2 report age and perceived difficulty as significant.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
8 Valid · 0 Not valid · 28 Not checked · 212 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“The findings revealed a significant reduction in brain connectivity and cognitive engagement in the LLM group, which showed the weakest activation across key neural networks associated with reasoning and memory.”
What the sources say
- Supportedcites [7] Kosmyna, N., Hauptmann, E., Yuan, Y., Situ, J., Liao, X.-H., Beresnitzky, A., Braunstein, I., and Maes, P. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. arXiv. source
The Abstract reports significantly different neural connectivity patterns across groups, with the LLM group showing the weakest overall coupling. The Related Work section links lower connectivity to reduced engagement of networks involved in reasoning and working memory. The paper measures connectivity rather than activation directly, but the citing sentence conveys its substantive finding. (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
Every source the paper cites for this claim was checked and supports it.
In the paper
“Participants who expressed high trust in AI also showed lower ownership of their final response, echoing findings by Kosmyna et al. [7] on reduced neural and textual authorship in AI-supported tasks.”
What the sources say
- Supportedcites Kosmyna et al. [7] Kosmyna, N., Hauptmann, E., Yuan, Y., Situ, J., Liao, X.-H., Beresnitzky, A., Braunstein, I., and Maes, P. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. arXiv. source
The citation is attached to the comparison with Kosmyna et al.’s findings, not to the citing authors’ observation about high trust in AI. The source’s Discussion reports reduced neural convergence and fragmented authorship in the LLM group, supporting that comparison in the essay-writing setting. (LLM, conf=0.90)
plainstates an association- 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.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Such patterns can erode the perceived value of human input, an observation echoed in the findings of Lee et al. [8]”
What the sources say
- Partly supportedcites [8] Lee, H.-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., and Wilson, N. (May, January 26). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects from a Survey of Knowledge Workers. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, Yokohama, Japan. sourceExperimental — accuracy not yet measured
Section 4.3.2 reports that trust in GenAI can discourage critical reflection and that some workers doubted their own abilities and accepted AI outputs by default. This supports a reduced perceived need for their own input, but the paper does not establish that unreflective use erodes the perceived value of human input more generally or over time. Page numbers were not available in our copy. (LLM, conf=0.82)
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
“Results show that unguided AI use fosters cognitive offloading without improving reasoning quality, whereas structured prompting significantly reduces offloading and enhances both critical reasoning and reflective engagement.”
boostedstates a causen = 150- 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
“Results show that unguided AI use fosters cognitive offloading without improving reasoning quality, whereas structured prompting significantly reduces offloading and enhances both critical reasoning and reflective engagement.”
plainstates a causen = 150- 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.
133 more claims
Create a free account to see all 248 claims and to verify the 4 references still open.
- Probably not a substantial claim. 52
- It may state two separate claims as one. 78
- The quote only partly states this claim. 2
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
§ 2. Materials and Methods
“Participants were randomly assigned to one of four conditions:”
§ 2.1. Participants
“All participants completed three experimental conditions (human-only, human + AI unguided, and human + AI guided).”
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
§ 2.2. Data Collection Procedure
“the order of the three human-involved conditions was counterbalanced across participants”
§ 2.2. Data Collection Procedure
“All participants first completed the task in the human-only condition, without access to AI. In the second condition (human + AI unguided), participants were given access to ChatGPT but were not provided with any guidance on how to use it. In the third condition (human + AI guided), participants received structured training designed to promote deliberate engagement and reduce cognitive offloading.”
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
§ 3.4.1. Predictors of Critical Thinking Performance
“Age, offloading, and perceived difficulty did not remain significant in the full model, likely due to multicollinearity or mediation effects.”
§ 3.4.1. Predictors of Critical Thinking Performance
“| Perceived Difficulty (Composite) | 1.8683 | 0.8281 | 2.2561 | 0.0248 | 0.2382 | 3.4984 | | Age | 1.8868 | 0.3175 | 5.9433 | <0.001 | 1.2619 | 2.5117 |”
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
§ 3.4.1. Predictors of Critical Thinking Performance
“perceived reflective engagement had a robust positive effect ( β = 1.12, p < 0.0001).”
§ 3.4.1. Predictors of Critical Thinking Performance
“| Perceived reflective engagement (Composite) | 0.3163 | 0.7366 | 0.4294 | 0.6679 | - 1.1336 | 1.7662 |”
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
§ 3.1. Overall Performance Differences Across Conditions
“While generative AI has the potential to enhance argument quality, this effect only materialises when users are guided to engage critically with the tool, thereby avoiding passive cognitive offloading.”
§ 3.1. Overall Performance Differences Across Conditions
“Those using AI without guidance also scored significantly higher than participants working without AI (mean difference = 2.56, p < 0.0001), though the improvement was modest.”
4 more findings
Create a free account to see every finding of this check.
Citations
Cites
Becoming a Self-Regulated Learner: An Overview
Zimmerman, B. J. (2002) · Theory Into Practice · doi:10.1207/s15430421tip4102_2 · Cited by 10,197 · cited 2×
“In the learning sciences, self-regulation has long been identified as a core determinant of deep engagement and transfer [5].”
Sweller, J., Ayres, P., & Kalyuga, S. (2011) · doi:10.1007/978-1-4419-8126-4 · Cited by 5,531
Sweller, J. (2011) · Psychology of Learning and Motivation · doi:10.1016/b978-0-12-387691-1.00002-8 · Cited by 4,856
Sweller, J. (2012) · Encyclopedia of the Sciences of Learning · doi:10.1007/978-1-4419-1428-6_446 · Cited by 4,405
Dual-Process Theories of Higher Cognition: Advancing the Debate
Evans, J. S. B. T., & Stanovich, K. E. (2013) · Perspectives on Psychological Science · doi:10.1177/1745691612460685 · Cited by 4,271
Things That Make Us Smart: Defending Human Attributes in the Age of the Machine
Robinson, R., & Norman, D. (1994) · Design Issues · doi:10.2307/1511661 · Cited by 1,528
AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking
Gerlich, M. (2025) · Societies · doi:10.3390/soc15010006 · Cited by 1,248
Self-citationAI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking
Gerlich, M. (2025) · doi:10.2139/ssrn.5082524 · Cited by 1,238
Self-citationLee, H.-P. (., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025) · Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713778 · Cited by 490 · cited 5×
“Such patterns can erode the perceived value of human input, an observation echoed in the findings of Lee et al. [8], who noted that many workers reported growing reliance on AI tools while showing decreased confidence in their own analytical contributions.”
InfluentialKosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X.-H., Beresnitzky, A. V., Braunstein, I., & Maes, P. (2025) · arXiv · doi:10.48550/arxiv.2506.08872 · Cited by 220 · cited 9×
“Kosmyna et al. [7], by contrast, used a longitudinal EEG-based design to monitor neural activation across repeated writing sessions, demonstrating reduced cognitive engagement among LLM users compared with control groups using search engines or unaided writing.”
InfluentialExploring Motivators for Trust in the Dichotomy of Human—AI Trust Dynamics
Gerlich, M. (2024) · Social Sciences · doi:10.3390/socsci13050251 · Cited by 46 · cited 3×
“Gerlich [19] noted this same dichotomy in the trust dynamic between humans and AI, where functional trust is linked to critical oversight, and blind trust to cognitive surrender.”
InfluentialSelf-citationArtelt, C., & Schneider, W. (2015) · Teachers College Record The Voice of Scholarship in Education · doi:10.1177/016146811511700109 · Cited by 35
The Role of Heuristics and Biases during Complex Choices with an AI Teammate
Gurney, N., Miller, J. H., & Pynadath, D. V. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i5.25741 · Cited by 10
Tang, K., Cooper, G., Rappa, N. A., et al. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4722537 · Cited by 5
Research works printed but not linked: 11
- Baasch, 2023, Who owns the thought? Epistemic responsibility in human–AI collaboration, Ethics Inf. Technol., 25, 39
- Rahimi, 2022, Automation and metacognition: Rethinking learner agency in AI-supported learning environments, Br. J. Educ. Technol., 53, 1124
- Schmid, 2023, Predictive text and learning performance: When writing feels easier, do we think less?, Comput. Educ., 191, 104649
- Liu, 2024, Cognitive Load in AI-Supported Learning: Rethinking Scaffolding through Prompt Design, Learn. Instr., 85, 101782
- Schluter, 2021, Prompt Literacy and Human Autonomy: Rethinking Competence in the Age of Language Models, AI Ethics, 2, 601
- Facione, P.A. (1990). Critical Thinking: A Statement of Expert Consensus for Purposes of Educational Assessment and Instruction (The Delphi Report), The California Academic Press for the American Philosophical Association.
- Halpern, D.F. (2013). Thought and Knowledge: An Introduction to Critical Thinking, Psychology Press. [5th ed.].
- Mata, 2012, Cognitive aging and decision making, Trends Cogn. Sci., 16, 261
- Yeo, 2023, The Persuasive Power of AI: Perceived Credibility, Bias Blindness, and Political Reasoning, AI Soc., 38, 433
- Dwivedi, 2023, Generative AI in practice: How knowledge workers use ChatGPT, Int. J. Inf. Manag., 71, 102653
- Leufer, 2022, Trustworthy AI in the Workplace: Between Functional Assistance and Cognitive Delegation, AI Ethics, 2, 439
Cited by
AI-overdependence and human cognitive decline: Hazards, evidence, and mitigation strategies
Noorbehbahani, F., & Oyibo, K. (2026) · Computers in Human Behavior Reports · doi:10.1016/j.chbr.2026.101102 · Cited by 12
Wang, T., Li, S., Li, S., Zhang, J., & Gao, J. (2026) · British Journal of Educational Technology · doi:10.1111/bjet.70059 · Cited by 14
Beşaltı, M., Kudubeş, A. A., Kudubeş, İ. E., Özbay, Ö., Sarıkahya, S. D., & Özbay, S. Ç. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2652063 · Cited by 15
Dong, A., Li, P., Chen, Y., Gibson, S., Zhao, L., & He, M. (2026) · Sustainability · doi:10.3390/su18115313 · Cited by 9
Fan, W., Cheng, L., Wang, Y., Zhao, Q., & Li, Y. (2026) · Behavioral Sciences · doi:10.3390/bs16061014 · Cited by 10
“On the other hand, cognitive offloading is associated with risks such as the deterioration of inherent cognitive abilities and weakened critical thinking arising from excessive reliance on external tools (Gerlich, 2025a, 2025b; Malik et al., 2023; Oakley et al., 2025).”
Influential
Metadata
- Authors
- Gerlich, M.
- Affiliations
- SBS Swiss Business School (CH)
- Corresponding author
- Gerlich, Michael
- ORCID
- Gerlich, Michael
- Venue
- Data
- Publisher
- MDPI AG
- Published
- 2025-10-30
- Volume
- 10
- Issue
- 11
- Pages
- 172
- ISSN
- 2306-5729 (ISSN-L)
- OpenAlex
- W4415720048
- Cited by
- 40
- References
- 7
- Citations by year
- 2026: 40
- FWCI
- 41.61
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Data (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Human Factors and Ergonomics › Human-Automation Interaction and Safety
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- generative AI, structured prompting, cognitive offloading, critical reasoning, reflective engagement, human-AI collaboration, AI governance, human-centered AI, enterprise AI adoption, cross-national comparison
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 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (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.