The Impact of Generative AI on Critical Thinking [pdf]

Automation, Atrophy, and Historical Parallels

• Many see the findings as unsurprising: any automation that removes practice opportunities weakens skills, echoing older “ironies of automation” work and long‑observed bank/office automation trends.
• Analogies are drawn to calculators, GPS, and physical labor: we gained efficiency but lost everyday arithmetic, navigation, and farm strength.
• Others stress important differences: losing mental math is minor compared to losing the ability to reason about systems, write clear code, or evaluate risk.

Search Engines vs LLMs

• One camp equates LLMs with Google: both make knowledge recall optional, so humans naturally offload.
• Critics argue LLMs are more dangerous: search at least forced people to compare sources, whereas LLMs “spoon‑feed” answers, making laziness and uncritical acceptance easier.

Software Engineering Skills and “Vibe Coding”

• Multiple anecdotes of engineers pasting stack traces or shell problems into LLMs and not reading the underlying error, feeling real skill atrophy.
• Concerns that juniors may never build fundamentals if they start with codegen tools; seniors fear losing sharpness needed for debugging, interviews, and architecture.
• Others say this is just moving up the abstraction ladder (like assembly → C), but skeptics note compilers are deterministic and reliable in ways LLMs are not.

Uses as Cognitive Amplifier or Gym

• Some report genuine cognitive benefits: language practice, better search over vague ideas, fast translation, and guided exploration of complex topics.
• A pattern emerges: experienced people with solid fundamentals feel amplified; novices risk skipping the learning necessary to benefit.

Education, Youth, and Assessment

• Several comments warn students: if AI does the work, your “own neural network remains untrained,” even if grades improve.
• Teachers describe strong grade pressure and low detection risk pushing honest students toward AI.
• Debate over whether AI should be used in K‑12 at all, given likely long‑term skill erosion.

Work, Management, and Skill Maintenance

• Delegating to AI is compared to managers delegating to staff: deep hands‑on ability tends to decay while higher‑level “specification” skills grow.
• Some propose formal “maintenance” of automatable skills (periodic exams, dedicated practice time) but doubt employers will sacrifice short‑term gains.

Methodology and Media Framing Concerns

• Several point out the study relies on self‑reported recollections of AI use, limiting its strength.
• The popular article is criticized as clickbait for pulling dramatic phrases from the introduction and older literature rather than the paper’s actual results.