OpenAI Sets Moonshot Goal of AI Superalignment in 4 Years
Yesterday OpenAI announced a major effort for what they call 'Superalignment' to align superintelligent AI with human intent and values. The company is dedicating 20% of its compute resources to the project and its being led by their chief scientist and co-founder. Before that on the Brief: Over the weekend, Google updated its policies to basically say that it's allowed to train AI on any public data we put online. At the same time, platforms like Twitter and Reddit are racing to keep other comp