Hacker News
A Developer Argues LLMs Are Costing Him the Savviness He Built From Debugging His Own Mistakes
Paolo Galeone wrote on August 29 that the loop that produced his competence — hours and sleepless nights on a problem, then resolving it himself — is gone now that he can have an idea and 'ask the model-of-the-month to implement it.' His specific claim is narrower than the usual deskilling complaint: he still spots the errors because of prior experience, but because the model supplies the correction, he does not learn from them, so the existing expertise depreciates without being replenished. The post ran 59 points and 75 comments and offers no measurements, only the mechanism.
↳ Follow the thread