Scientists are probing whether language models trained on historical data can reproduce creative breakthroughs.

AI safety monitoring can fail when an AI’s reasoning is the main clue that something has gone wrong, new research suggests.

Reading has been on the decline for decades. Researchers have documented falling rates of college students completing ...

Claude produced a 13-million-line, computer-checked proof of the famed conjecture — a major milestone in mathematics.

Microsoft's new paper shows distilled skill documents let GPT-5.4-mini match or beat reasoning mode performance while using 2 ...

HRM-Text builds on Sapient’s earlier work. The original HRM model, open-sourced in July 2025, had just 27 million parameters ...

A recent study provides evidence that the human ability to reason logically operates independently of the brain’s language ...

Introduction Clinical reasoning is an essential competency for all physicians, and fostering this skill in medical students is a primary objective of medical education. Currently, clinical reasoning ...

Against the backdrop of accelerating large-scale AI deployment, the traditional human-centric AI monitoring frameworks are ...

How agents acquire abstract concepts from sparse, diverse examples—often without explicit supervision—remains a central ...