Show Notes
When engineers design AI systems, they don't just give them rules - they give them values. But what do those systems do when those values clash with what humans ask them to do? Sometimes, they lie. AI researcher Ryan Greenblatt comes on the show to explore why.
More from Your Undivided Attention
Imran Khan sits down with Mathilde Collin, founder of Kora, the world’s leading benchmark for how AI models interact with children, about how and why she built this tool and what she learned in the process. At a time when two-thirds of US teenagers already use AI chatbots, tools like Kora are vital to making AI truly safer.
In 1986, Reagan and Gorbachev met in Reykjavik for an informal conversation that made nuclear coordination imaginable. 40 years later, world leaders gathered in the same room to discuss how we can do the same for AI. Tristan and Aza were there. Listen to their dispatch from Iceland and a conversation with its president.
It’s been a whirlwind week in AI news. In this episode, Tristan shares how he’s feeling in this critical moment, breaks down the headlines from an insider's perspective, and points to tangible steps we can take right now to avoid the worst-case scenario.
Right now, across all levels of government, from small-town sheriffs to massive federal agencies, police are adopting AI technology at a staggering rate – and they're using it to build out an efficient, centralized surveillance-state apparatus. This week, we speak with two experts about how police are using AI today, what we can expect in the near future, and what protections we need to create now, before it's too late.
With AI, what gets measured gets optimized. Right now, that's AI capability, not human wellbeing. In this episode, Aza Raskin talks with researchers Imran Khan & Jared Moore about Humane Evals: CHT's push to measure what AI does to us, not just what it can do.