Latest / Tech Unfiltered with Dr. Mike
OpenAI's Safety Warning: When Should AI Slow Down?
OpenAI's chief scientist argues that AI development may need restraint when safeguards cannot keep pace. Dr. Mike explains the September 6 essay and September 7 reporting, separating a proposed slowdown from an announced shutdown. We unpack alignment versus monitoring, OpenAI's reports about GPT-6 Astra and AI-assisted research, and why better performance does not automatically mean easier oversight. Company evaluations and forecasts are attributed as such; adversarial test results are not presented as evidence of deception in every ordinary chat. TAKEAWAYS - A warning and a forecast are not…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.