AI is transforming every aspect of life, but one very important area that I think we’re just starting to explore is how AI can speed up physical research. Specifically, you’ll hear the term “agentic AI” which basically means that the AI does more than just answer questions. It can, for example, execute some code to run an experiment or click buttons for you in various apps. All of that can sound vague, but a recent presentation about the new “Microsoft Discovery” platform is a solid example of this.
Before we get into that, have you heard of “PFAS” or maybe “forever chemicals”? Derek from Veritasium did a comprehensive overview of the problem recently.
My area of the company was exploring some extremely promising fluids that would revolutionize the way we cool datacenter computers, and while they were different enough to be safe from PFAS concerns, we stopped using them anyway.
But the problem still exists. How can we meet the ever-increasing demands for data center cooling? Enter the Microsoft Discovery platform. Using this AI tool, researchers were able to identify a new environmentally friendly coolant in just 200 hours instead of the years that it would normally have taken. Watch this 5 minute demo to see how it all worked:
It’s difficult to wrap our minds around the scale of progress here, but try to imagine this level of capability increase across ALL AREAS OF SCIENCE. These initially discoveries are just the first hints of the flood that is coming.