#58743
Mentioned in 1 episodes
Pragmatic AI Safety
A Framework for AI Safety Research
Book • 2023
Pragmatic AI Safety (PAIS) is a research approach that emphasizes performing impactful AI safety research without simultaneously advancing AI capabilities.
It highlights the importance of sociotechnical factors, such as safety culture, and encourages a systems view of AI safety to address complex interactions within the AI research ecosystem.
It highlights the importance of sociotechnical factors, such as safety culture, and encourages a systems view of AI safety to address complex interactions within the AI research ecosystem.
Mentioned by
Mentioned in 1 episodes
Mentioned by ![undefined]()
Arthur Conmy
29 snips
E48: Mechanizing Mechanistic Interpretability with Arthur Conmy
