Unlocking the Potential of AI in Science

Today, AI-enabled advances are revolutionizing biology, transforming science, and making our roads safer. However, this is only the beginning. If we fully capitalize on this opportunity, we can usher in a new era of discovery, empowering scientists across disciplines to solve challenges once thought unsolvable at a speed once deemed impossible.

The Opportunity to Advance Science in the AI Era

AI has already begun to enable landmark advances in science, dramatically accelerating the scientific process and allowing scientists to look at many things in new ways simultaneously. AI is also making it possible for many more people to participate in research. For example, AlphaFold alone has been accessed by 2.5 million researchers across 190 different countries. We’ve also made many of our landmark, AI-powered advances in connectomics, pangenome, weather, materials science, and climate models widely available to scientists.

Realizing the Potential of AI in Science Requires More Than Just Technological Breakthroughs

But realizing this immense potential of AI in science requires more than just technological breakthroughs; it demands a concerted effort to build the foundation for continued progress. Which is why countries that want to lead here need to work together to put in place the infrastructure, investments, and legal frameworks that support scientists, engineers, and a culture of ongoing innovation.

The Three I’s of Science in the AI Era

To give policymakers immediate, actionable steps, today we’re releasing our Policy Framework for Building the Future of Science with AI. The framework is based on three key principles:

• Infrastructure — Increase access to AI infrastructure: Most scientists won’t need to train their own large AI model, but they will need access to resources to fine-tune large models, run simulations to generate high-quality data, or train smaller AI models on their specialized data. Governments can achieve this by setting up National AI for Science Resource Centers, similar in concept to the U.S. National AI Research Resource.
• Investment — Invest in the science of AI: Groundbreaking scientific discoveries often require long-term commitment and sustained investment. Governments should create a list of priority areas to direct their funding and incentivize research collaboration through public challenges aimed at solving the most pressing issues.
• Innovation — Implement pro-science and pro-innovation legal frameworks: With global AI competition accelerating, we need to support innovation while establishing frameworks for high-risk applications. Regulatory uncertainty slows innovation and creates barriers for scientists and private investors. To address this issue, governments should establish pro-innovation regulatory regimes that support responsible and reasonable use of data, flexible copyright frameworks, and harmonized data privacy laws.

Accelerating Scientific Progress

With the right policy and investment frameworks, governments can help accelerate scientific progress by clearing the way for scientists to continue to deliver the kinds of breakthroughs that will power a brighter future for people everywhere. There are many more challenges out there for AI to solve — and many ways for countries to work together to promote major AI-led breakthroughs.