Google DeepMind has invested closely in scientific AI for years, and it paid off in 2024 when Demis Hassabis and John Jumper, the corporate’s CEO and director, gained the Nobel Prize in chemistry for AlphaFold, a specialised system that may predict the three-dimensional construction of a protein.
Now its rivals are working to catch up. In October 2025, OpenAI launched a group dedicated to AI for science, and Anthropic introduced a number of Claude options geared towards the organic sciences across the similar time. OpenAI specifically has known as constructing an autonomous researcher its “North Star.” It simply introduced GPT‑Rosalind, the primary in a deliberate sequence of specialised scientific fashions. Google launched its personal AI co-scientist software final February.
Below the hood, many of those AI-for-science methods are in actual fact a number of specialised AI brokers working in live performance. Google’s co-scientist makes use of a supervisor agent, a era agent, and a rating agent, amongst a number of others, with the intention to generate potential hypotheses and analysis plans in response to a purpose supplied by a human scientist. Extra just lately, researchers at Stanford’s AI for Science Lab, led by James Zou, devised a “digital lab” made up of brokers that took on the roles of specialists in several scientific fields. They discovered that their system might design new antibody fragments that bind to SARS-CoV-2, the virus that causes covid.
