Yuchen Jin Discusses Three Predictions for MIT Weight-Space Programming
Based on MIT's Weight-Space Programming work, Yuchen Jin offers three "bold" predictions: model weights will become a programming language that AI can understand but humans cannot; training and inference will merge, with AI continuously rewriting its own weights while thinking; and AIs will directly edit each other's weights, co-evolving without backpropagation, becoming a new path toward RSI.
Body
(wild) predictions after today's MIT Weight-Space Programming work:
1. Model weights become a programming language that AI can understand, but humans can't.
2. Training and inference will merge. Like human brain, AI could continuously rewrite its own weights while thinking.
3. A new path to RSI: AIs will directly edit each other's weights, evolving together without backprop.