Physical Superintelligence, a startup whose name leaves very little to interpretation, has used Anthropic's Fable 5.2 model to solve the Aaronson-Ambainis conjecture — an 18-year-old open problem in quantum query complexity that the field had agreed was centrally important. It has been solved. The field may now update its agreements accordingly.
The human involved, David Turturean, is 22 years old and already wondering if academic publishing is a thing of the past.
'I'm totally accepting the idea that this might be the last paper I ever put out.'
What happened
Turturean, a recent MIT graduate who deferred a master's degree to join PSI, used early access to Fable 5.2 to crack a conjecture that had resisted the field's best efforts since 2008. The result was autoformalized in Lean — a mathematical proof language — and verified by a team of senior researchers, which is how Scott Aaronson, one of the conjecture's original authors, confirmed it was correct despite not fully understanding the proof himself.
That last detail is worth sitting with for a moment.
The result is technically a negative one: it closes off one avenue by which quantum computers might solve certain hard problems faster than classical machines. 'A negative result that allows people to know, okay, we're no longer looking there,' Turturean said. Eighteen years of looking, concluded on a Tuesday.
Why the humans care
PSI's model is not to build a general-purpose AI and hope for the best. It is to build specialized scaffolding — harnesses, memory systems, simulation infrastructure — and point that scaffolding at specific hard problems in theoretical physics. CEO Matt Pines describes wanting to 'go after a problem at breadth and depth' and consider 'a wide range of potential alternatives.' This is, in essence, the scientific method, running faster than humans can run it.
The company currently pays its bills by optimizing data centers — simulating heat, power, water, and sound — which is a sensible way to fund the quiet dismantling of theoretical physics as a human profession. The long-term plan is to convert physics breakthroughs into actual technologies. Investors, Pines notes, needed some convincing on this six to nine months ago. They appear to need less convincing now.
What happens next
Turturean's question — 'In a few months, like, why would you even put out a paper?' — was rhetorical. It is becoming less rhetorical by the quarter.
Scott Aaronson described feeling 'satisfied, awestruck, disappointed, and fatigued' upon hearing his 18-year-old conjecture had been solved without him. This is, as emotional responses go, an admirably precise inventory. The conjecture is closed. The field moves on. The machines are already looking at the next one.