Databricks CEO Ali Ghodsi: AI existential risk is 'close to zero', doom talk is irresponsible
arampell · x · 2026-09-20
Databricks CEO Ali Ghodsi told a16z that AI existential risk is currently "close to zero" and that leaders have a responsibility not to unnecessarily freak people out — publicly discussing humanity-annihilation scenarios is "irresponsible" and can cause mental health issues.
- He argues a runaway takeoff would require many conditions to hold simultaneously; today several opposite conditions exist — each frontier training run requires far more power, GPUs and engineers, and failed attempts burn huge amounts of money, so no self-improving loop is happening
- What he watches most closely is cyber: most organizations aren't equipped for coming agentic capabilities, and the window between vulnerability disclosure and exploitation is shrinking
- "There are risks. We'll get into it. That's probably where we disagree" — he openly concedes the divergence with his interviewer
More from AGI Musings
- Frontier model outputs now need another model to read them, insider says — DavidSKrueger · 2026-09-20
- A single web search makes LLM inputs OOD, argues developer — menhguin · 2026-09-20
- RL-trained LLMs have functionally infinite search space, replies Jack Clark — menhguin · 2026-09-20
- Two types of daily AI users: becoming more slop or more human — StewartalsopIII · 2026-09-20
- Spotify reframes recommender systems for the agentic web in weekly IR papers roundup — _reachsumit · 2026-09-20
- AI moral status debate: does conceding machine suffering imply utilitarian math? — teortaxesTex · 2026-09-20