Designing Against Deskilling: Metacognitive Feedback Reduces Cognitive Offloading to LLM Assistants
Preregistered study of 704 participants shows metacognitive feedback reduces cognitive offloading to LLM assistants and improves unaided test performance.
A preregistered 2x2 online experiment with 704 participants plus a no-AI control tested two interventions against cognitive offloading during fraction-arithmetic practice with an LLM-based assistant. Metacognitive feedback that made offloading implications explicit reduced answer offloading (OR=0.47) and improved unaided test performance (OR=1.51). An effort-based reward that incentivized less assistance showed no measurable effect on either outcome. The authors position metacognitive feedback as a design lever to mitigate AI-induced deskilling.
- Metacognitive feedback cut answer offloading with odds ratio 0.47
- Same intervention improved unaided test performance with OR 1.51
- Effort-based reward showed no significant effect on outcomes
- Preregistered 2x2 design with 704 participants and no-AI control
From the original article
1 min read at arxiv.orgCognitive offloading to AI can reduce opportunities to practice skills, creating risks of deskilling. However, it remains unclear how to prevent deskilling without restricting access to AI.
By Sebastian Maier · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.