🧠 AI Ed Wiki

📄 Research Article

Is AI making us stupid?

● high · 2026-07-19cognitive-offloading, over-reliance, generative-ai, metacognition, learning-gains, ai-literacy, skill-decay, self-regulated-learning, rag, higher-ed

Summary

A 3-page perspective (opinion/review, not an empirical study) addressing whether AI use erodes human cognition. The authors' answer: not inherently — but the risk is real and follows the cognitive-psychology principle of cognitive offloading. When people delegate reasoning, writing, memory, or problem-solving to AI, they forgo the mental practice that builds and maintains those capacities. The threat is use-dependent, not intrinsic to the technology: AI that augments thinking preserves the underlying processes; AI that replaces them outsources exactly the practice that builds expertise.

Core distinction: skills vs. basic cognitive abilities

The article's organizing framework separates two facets of our cognitive systems:

  • Skills — learned, domain-specific behaviors supported by knowledge (arithmetic, flying, diagnosis, writing, programming). Acquired and maintained through practice [4]. Offloading practice to AI "will almost certainly compromise skill acquisition" [5] and can cause skill decay [5,7].
  • Basic cognitive abilities — foundational, domain-general capacities (working memory, selective attention) that underlie skills. May be more resilient to erosion: cognitive-training research shows gains are highly task-specific rather than broad [9], so basic abilities appear "stubbornly resistant to substantial change." Open question whether long-term or developmental-stage offloading could still shift them [11,12].

    • Evidence cited

  • Math learning (high school) [6]: Students with an AI that let them fully offload solving practice problems scored higher on practice but performed worse on a later no-AI test than students who never had AI. A third condition — a custom AI tutor that probed knowledge and filled gaps (rather than giving answers) — performed no worse than no-AI students, indicating the targeted skill was still acquired.
  • Endoscopy skill decay [8]: After an AI detection tool was introduced, adenoma detection rates fell from 28.4% → 22.4% in cases where the AI was unavailable — evidence of deskilling when the tool was withdrawn.
  • Knowledge acquisition (Box 1) [2,3]: Offloading to an external store during learning reduces later retrieval; an AI-summary study found advice was "briefer, less unique, and rated less helpful," with learners spending less time and feeling less ownership — shallower depth of learning than web search.

    • The "how we use AI matters" argument

    Whether skills survive offloading depends on the form the offloading takes [13]:

  • Completely offloading the task ("student has the AI submit the answer") → harm.
  • Having the AI provide an explanation, a suggestion, or act as a collaborator giving feedback [14], or emulating a thoughtful tutor [15] → can preserve or even boost skill despite reduced effort ("Coach not crutch" [15]).
  • Staying "in the proverbial cognitive loop" mitigates costs; costs are likely limited to the specific skills offloaded, not basic abilities.

    • Concluding remarks & open questions (Box 2)

    "It is far too early to say with certainty" the long-term effects. Open questions: prolonged offloading over years/decades; developmental-stage effects (children?); decay as a function of initial skill level and skill type; refresher-training strategies; designing AI to discourage harmful offloading (policy levers); whether people can learn to strategically offload; impacts on metacognition and source-monitoring (misattributing AI output as one's own); effects on dispositions toward thinking.

    Notably, the article cites the PNAS guardrails paper (Bastani et al. 2025, Generative AI Guardrails Harm Learning) and the "Coach not crutch" preprint (Lira et al. 2025) — making the offloading/crutch mechanism a shared thread across the wiki.

    Why this matters for the wiki

  • The canonical cognitive-offloading citation Cognitive Offloading needed; this perspective synthesizes the mechanism and the skills-vs-basic-abilities resilience distinction.
  • Conceptual bookend to the empirical RCTs already in the wiki: Generative AI Guardrails Harm Learning (unguarded tutor cut exam scores via crutch/offloading), Generative AI Reduced Study Time Math (population "cognitive surrender"), and contrasts with AI Generated Feedback Higher Ed (well-architected AI feedback matched teachers — the "coach not crutch" design).
  • Reinforces Over Reliance, Metacognition (source-monitoring), Self Regulated Learning (deliberate vs. passive offloading), and AI Literacy (using AI to augment, not replace).

    • Connected Concepts

  • AI Literacy
  • Metacognition
  • Self Regulated Learning
  • Generative AI
  • Higher Ed
  • RAG

    • Connected Articles

  • AI Generated Feedback Higher Ed — Artificial intelligence and feedback in university education: effectiveness and student perceptions
  • Cognitive Offloading Speedup Illusion — Cognitive offloading and the speedup illusion in human-AI interaction
  • Efficiency Gain Illusion AI Overreliance — The efficiency-gain illusion: People underestimate the rate of AI use and overestimate its benefits on simple tasks
  • GenAI Performance Vs Learning — Distinguishing performance gains from learning when using generative AI
  • Generative AI Guardrails Harm Learning — Generative AI without guardrails can harm learning: Evidence from high school mathematics
  • Generative AI Reduced Study Time Math — Faster Completion, Less Learning: Generative AI Reduced Study Time on Math Problems and the Knowledge They Build
  • A4l Analytics Pipeline — Generalizing a Highly Configurable Analytics Pipeline to Replicate and Support Educational Research Across Multiple D...
  • Aaai2026 Prompting Literacy K12 — Learning to Use AI for Learning: Teaching Responsible Use of AI Chatbot to K-12 Students Through an AI Literacy Module
  • Academiclaw Student Agent Benchmark — AcademiClaw: When Students Set Challenges for AI Agents
  • Access Not Enough AI Tutoring 2026 — Access is Not Enough: Human Support Improves Engagement with AI Tutoring
  • Adapt Adaptive Lesson Plan Transformer — AdaPT: Adaptive Lesson Plan Transformer for Cross-Regional and Differentiated Instruction
  • Adaptive Pretesting Retention — Do Gains from Generative AI-Enabled Adaptive Pretesting Persist? Evidence from a Retention Study
  • Affective Text Wearable Student Health — A Formative Study of Brief Affective Text as a Complement to Wearable Sensing for Longitudinal Student Health Monitoring
  • Agency Gap AI Writing — The agency gap in AI-supported writing: how reactive and proactive agent designs shape multimodal reasoning
  • Agent Voice Accents K12 Group Learning — Exploring How Agent Voice Accents Shape Human-AI Collaboration in K-12 Group Learning
  • Agentic AI Education Scoping Review — Agentic AI in Education: A Scoping Review of Research Landscape, Capabilities, and the Frontier Agent Paradigm
  • Agentic AI Pedagogical Best Practice 2026 — Agentic AI and Pedagogical Best Practice: The Tension Between Automation and Learning
  • Agentic Education Coding — Agentic Education with AI Coding Assistants
  • Agentic Literacy Debt — Agentic Literacy Debt: A Structural Problem the AI Literacy Field Has Not Yet Named
  • Agents That Teach Incidental Learning — Agents That Teach: Designing Incidental Learning Back into AI-Assisted Software Development
  • AI Adoption Training Public Sector — The Main Barrier to AI Adoption in the Public Sector is Lack of Training
  • AI Adult Learning Guidelines Dis2026 — Guidelines for Designing AI Technologies to Support Adult Learning
  • AI Agents Constructive Conflict Design Education 2026 — Enacting Constructive Conflicts with AI Agents to Enhance Reconsideration among Novice Interaction Designers
  • AI Assessment Scale Reform — A bit of chaos and madness": The AI Assessment Scale and the work of assessment reform
  • AI Assistance Discretionary Feedback — AI Assistance for Discretionary Work: Increasing Feedback Provision in Higher Education

    • Citation

    Cash, T. N., Kelly, M. O., Macnamara, B. N., & Risko, E. F. (2026). Is AI making us stupid? Trends in Cognitive Sciences