Leading South African physicists and mathematicians said OpenAI’s claimed physics breakthrough does not put AI close to a theory of everything. They spoke to TechCentral about the company’s Yang-Mills result.
A theory of everything would join quantum mechanics with Einstein’s theory of gravity. Scientists call this goal quantum gravity. A fourth expert, George Ellis, raised doubts about the OpenAI claim itself.
“It’s hard to predict, given the pace at which AI is advancing, but I would say that this is highly unlikely in the near future,” said Kevin Goldstein, a physics professor at Wits University.
On 26 October, OpenAI announced results produced by AI in maths, computer science and physics. One was a proof that pure Yang-Mills theory exists. This is a simple form of the theory behind the strong nuclear force, which holds atomic nuclei together.
A review committee led by Nobel prize winner David Gross is checking the work. On 9 November, it said OpenAI’s code proves the theory exists under the standard problem statements used in physics and maths. The code was written in Lean, a language that lets computers check maths proofs.
The committee’s assessment, published a day later, said the work does not prove a mass gap. A mass gap means the lightest particle in the theory has a mass above zero. Proving both existence and the mass gap is one of the Clay Mathematics Institute’s Millennium Prize problems, worth US$1 million each.
Goldstein said the biggest problem for quantum gravity is testing it. He said the energy levels where quantum gravity matters are hard for scientists to reach in experiments.
He said the route to the Yang-Mills proof was already known, and OpenAI worked through much of the technical detail faster than humans could. “For quantum gravity, we are not sure what the correct path is,” he said.
Vishnu Jejjala, who holds a research chair in theoretical particle cosmology at Wits, agreed. “Regarding quantum gravity, AI helping is realistic; AI solving the problem, less so,” he said.
Jejjala said the Yang-Mills problem had a known theory, written success rules and a plan, and a computer could check the answer. “Quantum gravity has none of these things,” he said.
“Choosing the correct principles is a different task from proving a stated theorem, and nothing in this result shows whether AI can do it,” Jejjala said.
Jonathan Shock, interim director of the UCT AI Initiative, said AI is quickly making progress on very hard problems. But he said joining physics into one theory is quite different.
“One thing is that there isn’t, as far as I’m aware, an agreed-upon statement of what is to be proved,” Shock said. He said the Millennium Prize problems, by contrast, are very clearly set out.
Shock said that without more experimental data, AI could find many theories that fit what is known. There would be no way to tell them apart. “Physicists will trust such a theory when they can explain it to each other without the machine,” he said.
Ellis, an emeritus professor at UCT, co-wrote a book with Stephen Hawking. He said he was not an expert on what AI can do in physics. He pointed TechCentral to a blog post by Columbia University maths lecturer Peter Woit.
In the post, Woit called OpenAI’s Millennium Prize framing “extremely misleading”. He said the Clay institute’s official problem rules exclude this kind of existence proof unless other properties, such as the mass gap, are also shown.
Woit also criticised the committee for keeping a 1,147-page document explaining the proof private. The committee’s assessment does say the work meets the Wightman axioms, the standard rules for a quantum field theory.
Jejjala said he is hopeful about AI in the longer term. “The way AI is advancing, within the next five years, the best mathematicians and the best theoretical physicists in the world will be silicon,” he said.
Source: this article is based on reporting by TechCentral. Image: TechCentral. Written with the help of AI and published by the Tzaneen Voice Technology desk. See our Editorial Standards.
Werner Jacobs covers technology for Tzaneen Voice. Werner reports on mobile networks, data prices, internet access, smartphones, apps, AI, cybersecurity and South African tech companies. Each story explains what a new product, price change or policy means for ordinary users in South Africa.