Eric Schmidt shared his opinion on the impact of AI on scientific progress.
One example is Nvidia’s creation of a digital twin of the Earth. Earth-2 which uses weather predictions from FourCastNet , an AI model that processes tens of terabytes of Earth system data. The model is able to predict the weather for the next 2 weeks tens of thousands of times faster and more accurately than current forecasting methods.
Schmidt argues that AI can rewrite the scientific process, saving time and effort on routine work while spurring creative invention and discovery. AI is already transforming the way scientific reviews are conducted. Tools such as PaperQA and Elicit use Large Language Models (LLMs) to scan databases of articles and generate concise and precise summaries of existing literature.
Artificial intelligence can also expand the search for hypotheses and test them faster, helping to formulate stronger hypotheses. For example, AI can help create more promising drug candidates.
Ultimately, most science will be conducted in “self-driving labs” – automated robotic platforms combined with artificial intelligence. Such labs are already popping up at companies like Emerald Cloud Lab and Artificial, and even Argonne National Laboratory.
However, Schmidt emphasizes that the limitations and risks of artificial intelligence must be taken into account. AI is a powerful tool that allows people to achieve more with less: less time, less skills, less equipment. In other words, AI lowers the entry barrier to science for new scientists. But these capabilities make AI a dangerous weapon in the wrong hands. Smart, well-informed regulation is needed to prevent both intentional and unintentional misuse of AI.