Type
Research
Section
Education and Teaching
Why needed
In January 2026 Karpathy noted he was 'slowly starting to atrophy my ability to write code manually' as a direct consequence of ~80% AI-assisted coding. Eureka Labs is premised on AI tutors accelerating human learning — yet the research just completed establishes he sees AI assistance as a double-edged tool that degrades the skills it replaces. Does his documented skill-atrophy concern inform how Eureka Labs thinks about AI in education (e.g., when to withhold AI help to preserve student capability)? This gap sits at the intersection of his personal safety-adjacent observation and his core educational mission.
Generated from
Follow-up from Issue #53: Has Karpathy proposed any technical solutions for the near-term risks he identified?
Type
Research
Section
Education and Teaching
Why needed
In January 2026 Karpathy noted he was 'slowly starting to atrophy my ability to write code manually' as a direct consequence of ~80% AI-assisted coding. Eureka Labs is premised on AI tutors accelerating human learning — yet the research just completed establishes he sees AI assistance as a double-edged tool that degrades the skills it replaces. Does his documented skill-atrophy concern inform how Eureka Labs thinks about AI in education (e.g., when to withhold AI help to preserve student capability)? This gap sits at the intersection of his personal safety-adjacent observation and his core educational mission.
Generated from
Follow-up from Issue #53: Has Karpathy proposed any technical solutions for the near-term risks he identified?