LessWrong (30+ Karma) “AI safety undervalues founders” by Ryan Kidd
Nov 16, 2025
Ryan Kidd, an AI safety organization founder and field-builder, passionately argues that the ecosystem undervalues founders compared to researchers. He highlights how this status gradient stifles talent and slows growth in AI safety initiatives. Kidd discusses the multiplier effects of founders on recruiting and training talent, and how their traits clash with typical research incentives. He proposes actionable steps to elevate the importance of founders, including changing funding dynamics and supporting early-career entrants to ultimately boost the entire field.
AI Snips
Chapters
Transcript
Episode notes
Founders Create Multiplicative Capacity
- Founders and field-builders have multiplicative effects on recruiting, training, and deploying AI safety talent.
- Scaling the field requires multiplicative capacity like training programs and founders, not just marginal hires.
Founder Traits Clash With Research Culture
- The cognitive style that excels at AI-risk research often conflicts with founder traits like risk-taking and shipping.
- This anti-correlation causes selection pressure away from founding within the AI safety community.
Incentives Favor Research Over Building
- Incentive structures in AI safety tend to reward research prestige over organizational building.
- Higher status and compensation often flow to researchers, discouraging potential founders.