OpenAI talks GPT-6 Astra and Millenium Prize, researchers create WeWorm exploit & IBM’s US Open app
The discussion centers on OpenAI’s Astra demo, weighing its impressive multimodal capabilities and mathematics progress against concerns about computational cost, practical usefulness, and whether bigger models truly solve real-world agent problems or merely showcase flashy capabilities.
MAIN POINTS FROM TRANSCRIPT
- Astra’s demo impressed with 2D-to-3D rendering and task execution like ordering food.
- Olivia questioned whether such capabilities justify the massive token and compute usage.
- The panel debated AGI, noting no agreed-upon definition exists in the AI community.
- Speakers suggested current agent systems may need better world access, not just larger models.
TAKEAWAYS
- Model size alone may not be the best path to useful AI agents.
- Practical deployment matters as much as benchmark or demo performance.
- AGI claims remain ambiguous without a shared definition.
- Real progress may depend on improving how agents interact with the world.