Companies hype risks while hiding training details
tallinzen · x · 2026-08-28
Tal Linzen highlights a paradox: companies tout model capabilities and dangers while keeping training details secret. The public only gets a glimpse through public failures.
A quoted tweet adds that OpenAI likely uses multi-agent training where sub-agents delegate and communicate via official tool calls on the same task.
More from AGI Musings
- Are Agents Getting Smarter or Just Better Tooling? — sophiamia1346 · 2026-08-28
- Compute Will Move to Deserts, Oceans, and Orbit to Evade Local Blocks — Dr_Singularity · 2026-08-28
- Upcoming Interview: Worthy Successor & AGI Governance — danfaggella · 2026-08-28
- Viewpoint: Blocking Automation Maintains Scarcity and Suffering — TinfoilTricorn · 2026-08-28
- GTA 6 as the Last Pyramid: The End of Human-Heavy Game Development — genmon · 2026-08-28
- Shunyu Yao: RL finally works, marking the start of AI's 'Second Half' — dejavucoder · 2026-08-28