A former OpenAI researcher predicts that $100 billion will flow into training data because scaling alone won't be enough.

He argues that models are becoming more specialized in coding and math while stagnating or regressing in other areas.

AI-generated: summaries written by AI from the linked sources. How we use AI