recursive self-improvement

OpenAI Says AI Scaling Needs Stronger Safety Bars

OpenAI Chief Scientist Jakub Pachocki argues that rapid AI progress is making alignment and monitoring the central limits on responsible development.