The public is far from understanding the capabilities of the latest AI models, which are not publicly accessible

The field of preventing harm from advanced AI systems, spanning alignment, evaluation, and decisions to slow or withhold models whose capabilities outpace their safeguards.

Includes content about AI's impact on society, covering work, education, healthcare, culture, privacy, ethics, and everyday life.

Research topic (Topic).