Would You Say Stop? - AI risk in a world of non-experts

For most of us, the most dangerous thing we’ll do today is drive a car, hurtling a ton of metal at high speed down the road, surrounded by dozens of strangers doing the same, but somehow, most of the time, it ends without incident. That’s not because we’re great drivers, but rather because traffic is instead built assuming the exact opposite: that we all make mistakes constantly. Behind the everyday act of driving sits a massive system of risk management and safety design, and in this episode, I’ll explore why the future of agentic AI might become a system very much like traffic - and what we can learn from the roads about building safe AI Agent systems.