Inference Time Economics
Neel discusses the balance between inference time and training time compute, suggesting that future models may find a more equitable ratio. He emphasizes the importance of AI safety as human-level AI becomes technologically feasible, highlighting the need for extensive research in safety and interpretability to navigate potential risks and benefits.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Neel Nanda - Mechanistic Interpretability (Sparse Autoencoders)
Related Questions
Why is AI safety important in the context of the episode Neel Nanda - Mechanistic Interpretability (Sparse Autoencoders) and the clip Inference Time Economics?
Can we predict AI capabilities based on the episode Neel Nanda - Mechanistic Interpretability (Sparse Autoencoders) and the clip Inference Time Economics?
Can we predict AI capabilities?