AI Radar

Noteworthy AI conversations and writing, summarized from publisher notes.

Sam Altman on AGI, GPT-5, and What's Next

· OpenAI Podcast

In the inaugural episode of the OpenAI Podcast, host Andrew Maine interviews CEO Sam Altman on a wide range of topics. Altman reframes AGI as a moving target: existing models already exceed what most people would have called AGI five years ago, and each year more people will feel we've reached it as the bar keeps rising. He distinguishes AGI from superintelligence, which he defines as AI systems capable of autonomous scientific discovery—pointing to signs of progress but explicitly saying OpenAI has not 'figured it out' yet, though confidence in the direction is growing. On products, Altman says GPT-5 will likely arrive 'sometime this summer' and acknowledges the current model naming/numbering (o3, o4-mini, GPT-4.0, etc.) has become confusing. He hints at a cleaner post-GPT-5 versioning approach and praises memory, Deep Research, and Operator (with o3) as recent standouts. He also previews a hardware device being built with Jony Ive, though it won't ship for a while, and discusses the delicate question of how (or whether) to introduce advertising without compromising the integrity of model outputs. On infrastructure and policy, Altman describes Project Stargate—a roughly $500B effort to finance and build unprecedented compute capacity—as essential because demand massively outstrips what OpenAI can currently serve. He addresses energy needs (favoring an all-of-the-above mix with enthusiasm for advanced nuclear), criticizes Elon Musk for allegedly trying to derail OpenAI's UAE partnership, and acknowledges strong competition from Anthropic and Google's Gemini 2.5, saying the field is no longer zero-sum. He also addresses the sycophancy issue from a recent model update, the New York Times lawsuit over chat record preservation (which he calls an 'overreach' but hopes will accelerate a societal conversation about AI privacy), and offers advice to young workers: learn to use AI tools, and invest in resilience, adaptability, and creativity.

AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.