site - AI 2027
automatic AI researcher dim - takeoff speed -- fast
By this point “finishes training” is a bit of a misnomer; models are frequently updated to newer versions trained on additional data or partially re-trained to patch some weaknesses.18
One such breakthrough is augmenting the AI’s text-based scratchpad (chain of thought) with a higher-bandwidth thought process (diff - neuralese vs. COT). Another is a more scalable and efficient way to learn from the results of high-effort task solutions (iterated distillation and amplification).
Says they are trying to describe Median outcome