Simply this week, Pushmeet Kohli, Google Cloud’s chief scientist, printed a bit in a particular AI and science situation of the journal Daedaluswriting: “We’re shifting towards AI that doesn’t simply facilitate science however begins to do science.” With autonomous AI scientists on the horizon, it’s tougher to justify large efforts to develop super-specialized instruments—even one like AlphaFold, for which DeepMind scientists received a Nobel Prize, or a doubtlessly life-saving system like WeatherNext. It additionally heralds a far stranger future for science, by which people and AI techniques collaborate as friends—or AI even makes scientific progress by itself.
To be clear, Google doesn’t look like abandoning its work on specialised AI for science instruments. AlphaGenome and AlphaEarth Foundations, that are educated for genetics and Earth science functions respectively, have been launched final summer time, and the latest model of WeatherNext got here out in November.
What’s extra, such instruments stay extraordinarily fashionable amongst scientists. Final yr, as an example, Google reported that protein construction predictions from AlphaFold have been utilized by over three million researchers worldwide. And Isomorphic Labs, a Google subsidiary that goals to make use of AlphaFold and associated applied sciences to develop new medicine, simply raised a $2 billion Collection B funding spherical.
However there are concrete indicators of realignment, in each enthusiasm and sources. Final month, the Los Angeles Occasions reported that Google fellow John Jumper, who received the Nobel for AlphaFold, is now engaged on AI coding, not on science-specific AI instruments. It’s not shocking that Google is assigning its finest minds to the coding drawback, as the corporate has just lately taken a reputational hit as a result of its coding instruments don’t presently stand as much as these supplied by Anthropic and OpenAI. However it might additionally sign a prioritization of agentic science on Google’s half, as coding talents are key to the success of a few of these techniques.
Throughout the trade, agentic researcher techniques are exhibiting actual potential. This week, OpenAI introduced that one in every of their fashions had disproved an essential arithmetic conjecture—maybe probably the most significant contribution that generative AI has made to arithmetic to date, in line with some mathematicians.
Importantly, the mannequin utilized by OpenAI shouldn’t be specialised for fixing mathematical issues, and even for analysis; in line with the corporate, it’s a general-purpose reasoning mannequin within the vein of GPT-5.5. If common brokers could make unbiased contributions to mathematical analysis, they may quickly be capable of do the identical in science (although the truth that concepts in science have to be verified experimentally makes it a harder area for AI).
