US Institute Appoints “AI Doomer” to Head AI Safety

US AI Safety Institute Announces Leadership Amid Controversy

The US AI Safety Institute, nested within the National Institute of Standards and Technology (NIST), has unveiled its executive leadership team, marking an end to widespread speculation. At the helm of AI safety efforts is Paul Christiano, a notable figure formerly associated with OpenAI. Christiano brought to the fore a critical AI safety method known as reinforcement learning from human feedback (RLHF). However, his prediction that AI could catalyze a "doom" scenario with a 50 percent likelihood has sparked debate and concern.