Eric Schmidt: This is how AI is going to transform the way science is done

AI can also broaden the hypothesis-seeking network and narrow the network faster. As a result, AI tools can help make stronger hypotheses, such as models that spit out more promising candidates for new drugs. We’re already seeing simulations running multiple orders of magnitude faster than they did a few years ago, allowing scientists to try out more design options in the simulation before running experiments in the real world.

Caltech scientists, for example, used a AI fluid simulation model to automatically design a better catheter that prevents bacteria from moving upstream and causing infection. This kind of capability will fundamentally change the incremental process of scientific discovery, allowing researchers to design the optimal solution from the outset rather than progressing through a long line of progressively better designs, as we have seen in years of filament innovation in light bulb design. .

By moving to the experimental stage, AI will be able to conduct experiments faster, cheaper and on a larger scale. For example, we can build AI-powered machines with hundreds of micropipettes running day and night to create samples at a rate no human could match. Instead of limiting yourself to just six experiments, scientists can use AI tools to perform a thousand.

Scientists who are concerned about their next fellowship, publication, or commissioning process will no longer be constrained to safe experiments with the highest odds of success; they will be free to pursue bolder and more interdisciplinary hypotheses. When evaluating new molecules, for example, researchers tend to stick to candidates similar in structure to those we already know, but AI models don’t have to have the same biases and constraints.

Eventually, much of the science will be conducted in « self-driving laboratories, » automated robotic platforms combined with artificial intelligence. Here, we can bring AI prowess from the digital realm to the physical world. Such autonomous driving laboratories are already emerging in companies such as Emerald Cloud Lab AND Artificial and also to Argonne National Laboratory.

Finally, in the analysis and conclusion phase, self-driving laboratories will go beyond automation and, informed by the experimental results they have produced, will use the LLMs to interpret the results and recommend the next experiment to be performed. Then, as a partner in the research process, the AI ​​lab assistant could order supplies to replace those used in previous experiments, and set up and run the next recommended experiments overnight, with results ready to be delivered in the morning. all while the experimenter is at home sleeping. .

Possibilities and limits

Young researchers might fidget about their seats at the prospect. Fortunately, the new work emerging from this revolution is likely to be more inventive and less mindless than most current lab work.

AI tools can lower the barrier to entry for new scientists and open up opportunities for those traditionally excluded from the field. With LLMs able to assist in building code, STEM students will no longer need to master obscure programming languages, opening ivory tower doors to new, non-traditional talent and making it easier for scientists to engage in fields beyond their own . Soon, specially trained LLMs could go beyond offering first drafts of written work as grant proposals and could be developed to offer ‘peer’ reviews of new papers alongside human reviewers.