Country of Geniuses

AI agents at work

A country of geniuses, arriving one field at a time.

AI agents are moving from writing code to doing science: proving theorems, predicting molecular structures, and proposing experiments. This is a running record of where they are doing real work.

The fields

Recent milestones

  1. Jul 21, 2025

    Mathematics

    Two AI systems reach gold-medal standard at IMO 2025

    Models from Google DeepMind and OpenAI each solved five of six olympiad problems, writing natural-language proofs within the competition time limits.

    Google DeepMind

  2. May 14, 2025

    Mathematics

    AlphaEvolve improves on a 1969 matrix multiplication result

    An evolutionary coding agent found a way to multiply 4×4 complex matrices with 48 scalar multiplications, improving on Strassen's algorithm for this case.

    Google DeepMind

  3. Feb 24, 2025

    Software

    Claude Code launches as an agent in the terminal

    Anthropic released a coding agent that reads a codebase, edits files, and runs commands and tests on its own, working directly in the developer's terminal.

    Anthropic

  4. Feb 19, 2025

    Biology

    AI co-scientist generates and ranks research hypotheses

    Google introduced a multi-agent system that proposes, debates, and refines hypotheses. Some of its drug-repurposing suggestions were supported by laboratory experiments.

    Google Research

  5. Oct 9, 2024

    Biology

    AlphaFold's creators share the Nobel Prize in Chemistry

    Demis Hassabis and John Jumper were recognized for protein structure prediction, sharing the prize with David Baker for computational protein design.

    The Nobel Prize

  6. Jul 25, 2024

    Mathematics

    AlphaProof and AlphaGeometry 2 reach IMO silver-medal standard

    The systems solved four of six problems from the 2024 International Mathematical Olympiad, scoring 28 of 42 points.

    Google DeepMind

  7. Dec 14, 2023

    Mathematics

    FunSearch finds new constructions for the cap set problem

    A language model paired with an automated evaluator found larger cap sets than any previously known, a new result on an open problem in combinatorics.

    Nature

  8. Nov 29, 2023

    Chemistry & Materials

    GNoME predicts 2.2 million new crystal structures

    A graph neural network from Google DeepMind proposed 2.2 million new inorganic crystals, around 380,000 of them predicted to be stable.

    Google DeepMind