Large language models’ emergent abilities are improving with scale; as scale grows, where are LLMs heading? Insights from Ray Solomonoff’s theory of induction and stochastic realization theory may help us envision — and guide — the limits of scaling.Read More