Using AI for just 10 minutes erodes your ability to persist at hard things
A new study claiming that just 10 minutes of using AI tools can erode people’s persistence on hard tasks sparked sharp disagreement. Many commenters argue the experiment is poorly designed and simply shows short‑term “warm‑up” or motivation effects when a helpful tool is suddenly removed, likening AI to calculators, power tools, or search engines that shift effort rather than destroy ability. Others worry about longer‑term cognitive “fitness,” loss of foundational skills, and growing dependence on AI, while a sizable group reports feeling more productive, creative, and focused on higher‑level problems when they offload routine work to models.
Perceived Effects on Persistence & Cognition
- Some argue AI reduces tolerance for tedium, making it harder to stick with difficult tasks once AI is removed, similar to “learned helplessness.”
- Others see any short‑term drop in performance after AI is withdrawn as expected “warm‑up” loss, not long‑term cognitive damage.
- Several note that AI especially affects motivation: once you’ve had fast, assisted answers, going back to slow, manual work feels worse.
Tool Analogies & Historical Parallels
- Many compare AI to calculators, power tools, cars, TV, search engines, and word processors: tools that shift which skills are practiced.
- Critics respond that unlike saws or calculators, AI can take over higher‑level decisions, potentially atrophying judgment and problem analysis.
- Broader “domestication” analogies appear: loss of hunting skills vs gaining new “basic skills” like navigating supermarkets or information systems.
Personal Experiences: Productivity vs Dependence
- Several describe dramatic productivity gains: building compilers, OSes, websites, research outputs, or complex hardware+software systems with AI assistance.
- These users say AI lets them tackle more ambitious problems, “thinking things into existence,” but admit it requires intense focus to guide, review, and correct.
- Others report the opposite: feeling that AI coding becomes “putting square blocks in square holes,” or that overreliance makes non‑AI work feel daunting and empty.
Study Design & Critique of the Paper
- Multiple comments criticize the study’s design: extremely short time scales; no comparison groups using calculators, books, or human experts.
- Some suspect the experiment setup (removing AI mid‑task) mainly measures mindset switching and frustration, not deep, lasting erosion of ability.
- The article title is widely seen as overstated relative to the reported evidence.
Skill, Authorship & “Moving Up the Stack”
- There is debate over whether AI‑assisted work is “your” creation, likened variously to using compilers, hiring freelancers, or directing a film.
- Some fear that as AI handles more detail, long‑cultivated technical moats (e.g., deep coding ability) lose value; others welcome a shift to higher‑level orchestration.
- A few emphasize strict compartmentalization: AI for low‑value scaffolding, hand‑crafted code for core logic, to avoid ending up dependent on or buried in “AI slop.”