Quote · The a16z Show
Building the Physical AI Stack | Travis Kalanick on TBPN
Where this was said
AI Safety, Asimov, and the Market as Alignment Mechanism
At 24:44 · chapter starts 23:35
The hosts pivot to AI safety and Asimov's Three Laws of Robotics, asking whether Kalanick has ever had doubts about whether alignment will be solved. His answer is grounded in decades of entrepreneurial failure: the things he built that nobody liked failed. If AI builds something humans don't want, the same thing will happen. He notes that current AI models are arguably too eager to please — not a drift-toward-danger concern. His safety philosophy is less about Asimovian laws and more about incentive structures: robots are owned by someone with a bank account, who pays based on value delivered. Anti-human AI won't find buyers. Kalanick acknowledges there are real collision cases and prioritization dilemmas — the spirit of Asimov's laws — but he says rather than writing sci-fi about them, he's just building the thing and making sure the machine stays on the road.
Kalanick isn't worried about robot doom — he's pragmatic. If you build something anti-human, you won't find customers, and you won't make it. He notes that today's AIs are almost pathologically eager to please. His safety model: keep the machine on the road and make something people love.
