The reason 'thinking out loud' helps AI solve hard math is because it’s secretly turning one giant nightmare of a problem into a bunch of easy multiple-choice questions.
April 13, 2026
Original Paper
How does Chain of Thought decompose complex tasks?
arXiv · 2604.08872
The Takeaway
Researchers discovered a power law that explains exactly why 'thinking step-by-step' reduces error rates. It also identifies a point of diminishing returns, helping us predict when more reasoning steps stop being helpful and start wasting compute.
From the abstract
Many language tasks can be modeled as classification problems where a large language model (LLM) is given a prompt and selects one among many possible answers. We show that the classification error in such problems scales as a power law in the number of classes. This has a dramatic consequence: the prediction error can be reduced substantially by splitting the overall task into a sequence of smaller classification problems, each with the same number of classes ("degree"). This tree-structured de