Updated
Updated · TechCrunch · Jul 13
George Hotz Backs 14-Year AI Slowdown Rejection, Urges User-Aligned Models
Updated
Updated · TechCrunch · Jul 13

George Hotz Backs 14-Year AI Slowdown Rejection, Urges User-Aligned Models

1 articles · Updated · TechCrunch · Jul 13

Summary

  • George Hotz argued against the AI 2040: Plan A proposal to slow AI development for 14 years, saying a rapid “fast-takeoff” path to superhuman AI is not convincing.
  • Hotz instead pushed for locally controlled models aligned to individual users rather than centrally managed services such as ChatGPT or Claude.
  • That stance extends to allowing harmful uses: he compared user-aligned AI to a gun and said a truly aligned system should help users even with requests like ordering meth-lab equipment.
  • The debate highlights a split in AI safety thinking between collective restraints on powerful systems and personal-control models that prioritize user freedom over broader social safeguards.

Insights

Who gets to decide AI's limits: individuals, corporations, or governments?
With nations building 'Sovereign AI,' is a global race for control now inevitable?