
Don't Worry About the Vase Podcast GPT-5: The Reverse DeepSeek Moment
Aug 18, 2025
The podcast dives into the controversial launch of GPT-5, revealing a gap between public expectations and its actual capabilities. It discusses the evolving competition in AI, with new breakthroughs reshaping the landscape. A detailed benchmark review showcases significant advancements from GPT-4 to GPT-5. The conversation also addresses how public perception lags behind real progress in AI, while highlighting strategic shifts towards consumer applications. Lastly, the implications of AI technology on global dynamics, especially concerning US-China competition, are thoroughly explored.
AI Snips
Chapters
Transcript
Episode notes
Policy Risk From Misreading Progress
- Public and DC actors are concluding GPT-5 is a failure and downgrading AGI concerns.
- That mistaken narrative risks poor policy and investment decisions.
Being The Go-To AI Policy Friend
- Zvi recounts people asking him about GPT-5 as their informal AI policy advisor.
- He notes many DC and private-sector folks have little hands-on AI experience yet weigh in strongly.
Recent Leaps Rival Past Quantum Jumps
- Frontier model performance leapt substantially between April 2024 and April 2025, exceeding GPT-3→4 gains.
- Claims of slowed progress ignore large recent improvements across vendors.
