Startup Uses AI Model to Solve 18-Year-Old Quantum Computing Problem

A recent MIT graduate working at the startup lab Physical Superintelligence used an early version of an artificial intelligence model to solve an eighteen-year-old math problem. The problem, known as the Aaronson-Ambainis conjecture, deals with how much faster quantum computers can process certain tasks compared to traditional computers. One of the original researchers who posed the problem confirmed the proof appears correct after it was translated into a specialized computer language and reviewed by top experts.

Building a Business on Scientific Discovery

The startup aims to build a business by tackling complex scientific problems using custom software harnesses and simulation tools. While the company currently makes money by optimizing data centers for factors like power and heat, its ultimate goal is turning theoretical physics breakthroughs into real-world technology. Leaders at the firm believe that theoretical physics is rapidly shifting toward machine-led discovery.

The Challenges of Real-World Physics

Applying artificial intelligence to the physical world remains harder than solving abstract math problems because physical theories must be tested against reality. The startup is working on ways to quickly validate computational results with physical experiments. As automated scientific discoveries become more common, researchers and academics are still working to establish best practices and standards for sharing these rapid breakthroughs.