PSI claims its AI system has cracked 5 open quantum problems

Is physics about to be conquered by AI too? 5 open quantum problems that had stood for over 20 years have been cracked.

The one behind it is a mysterious company called Physical Superintelligence (PSI).

They claim that, with the help of their self-developed AI research system, they have achieved results on 5 open problems in quantum information theory, and that the paper, Lean formal proofs, and verification code have all been posted on GitHub for scholars to review.

So who exactly is this PSI, which was founded less than two months ago?

What did PSI solve?

The quantum information science that PSI focuses on this time mainly studies how to use the special laws of quantum mechanics to store, process, and transmit information, thereby breaking through the computational limits of traditional computers and exploring more secure ways of communication.

And the five difficult problems they challenged this time with AI all come from the 27-year-old Open Quantum Problems (OQP) catalog.

This open problems catalog was established in 1999 by the German physicist Reinhard Werner and is specifically dedicated to collecting important research problems that have long stumped the quantum information community.

The five problems this time span five core directions: entanglement purification, quantum field theory, high-dimensional quantum communication, quantum channel capacity, and quantum causality, including:

We will not expand on the specific proof process. It is enough to know that 3 of the 5 problems on which PSI made progress this time (OQP 2, OQP 12, OQP 27) were all included at the beginning of this century. That gives a sense of it.

"The world's most advanced physics research laboratory"

Clicking into PSI's website, it can simply be described as empty; cracking the quantum information problems this time is also their first result.

After all, this company only officially made its public debut on September 1 this year.

But it astonishes once it speaks: at its debut, it announced grand ambitions, claiming that it wants to build "the world's most advanced physics research laboratory."

PSI's founding team is rather interesting, and founder and chief scientist Alex Wissner-Gross is the most interesting of all.

This person once completed three undergraduate majors at MIT at the same time—physics, electrical engineering, and mathematics—and also graduated first in his class in the School of Engineering...

What does it mean to complete these three at the same time? Let me put it this way: after him, MIT abolished the option of "majoring in three fields at the same time," because "too few people can do this, and there is little point in offering the option."

Therefore, Wissner-Gross can even have the title of "the last undergraduate in MIT history to graduate with three majors."

Perhaps after testing the waters, the big shot felt that he still liked physics, so in 2007 he earned a PhD in physics at Harvard.

As early as 2013, when deep learning was still in its early stages, he took the TEDx stage and used the "causal entropic force" in physics to explain intelligent behavior.

He proposed a very bold idea.

So-called intelligence can be understood as a tendency to continuously increase the possibilities for future action.

This may be a bit hard to understand, so let me give an example.

For example, when you are squeezing onto the subway, why do you try hard to keep your balance and avoid falling?

From his theory's perspective, this is because once you fall, the actions you can choose are reduced. Staying on your feet means that in the future you can still move forward, backward, left, and right.

Then, as long as the rule of "maximizing future possibilities" is followed, the system may spontaneously exhibit intelligent behaviors such as tool use and cooperation.

Although it sounds reasonable, this idea has not yet been proven.

It is this brilliant Wissner-Gross who could no longer hold back this year and proposed to "Solve Everything."

In two articles published in February this year, he put forward the current PSI concept.

Among them are some flashes of insight that echo the present: "The singularity needs a physics department."

"The era of proof has arrived."

Besides this big shot, PSI's co-founder and president Alex Klokus is also very interesting; he is the founder of the well-known American tech media outlet Futurism.

Its financing background gives a glimpse of its strength. At its September debut, PSI announced that it had completed a $58 million seed round.

The lead investor is Breakthrough Energy Ventures, founded by Bill Gates.

Behind this fund are tech and business heavyweights such as Amazon founder Bezos, Alibaba founder Jack Ma, and Virgin Group founder Branson, and it specializes in investing in frontier technologies with the potential to drive energy transformation.

The follow-on investor lineup also includes the veteran Silicon Valley venture firm SV Angel. Its founder, Ron Conway, is a famous angel investor who in his early years backed tech companies such as Google, PayPal, and Twitter.

Not only that, former a16z partner Balaji Srinivasan also participated in the investment, and it also attracted individual investors from tech companies such as OpenAI, Nvidia, Hugging Face, and Oracle.

Let scientific discovery achieve "scaled production"

Having said so much, to summarize PSI's vision in one sentence, it is actually to make scientific discovery go "from genius to scale" (From Genius to Scale).

For centuries, the leaps in humanity's basic science have almost all relied on individual geniuses like Newton and Einstein, who had flashes of inspiration.

But PSI believes that this purely handcrafted, workshop-style scientific research has hit a ceiling. Otherwise, why are we in the 21st century still standing on the foundation of 20th-century physics?

Scientific discovery should no longer be constrained by human time.

PSI calls its basic research unit an Autonomous Campaign, hoping to keep research running continuously from problem definition, hypothesis generation, validation, and experimentation to engineering deployment.

PSI also believes that in the AI era of scientific research, although ideas are no longer scarce, truth is still scarce.

Large models can rapidly generate hundreds of conjectures, but these are still far from becoming truth.

To solve this problem, PSI's self-developed AI research system Emmy (a tribute to the great 20th-century mathematician and founder of theoretical physics Emmy Noether) separates the two steps of "generation" and "validation."

Its workflow is roughly like this.

First, when faced with a grand open problem in physics, Emmy first calls on frontier large models to break it down into countless tiny, verifiable mathematical sub-hypotheses.

Second, the hypotheses are thrown directly into the Lean formal proof assistant, symbolic computation engines, and high-precision validation code for rigorous screening. In this way, hypotheses that report errors are eliminated on the spot, while paths that pass local validation are retained and further derived.

As long as the final Lean code compiles, it means that this physical theorem is 100% impregnable in mathematical logic.

One More Thing

It sounds great, but here comes the question: how would a company that researches physics theories like this make a profit?

Today's supercomputing clusters are essentially extremely complex macroscopic physical systems, with highly coupled engineering optimization challenges among power supply, cooling, networking, and computing resources.

This is right up their alley. The Emmy system can start from underlying physical laws, treat power supply, heat dissipation, communication, and computing power as a single holistic equation to solve, and find a higher energy efficiency ratio from physical limits.

More interestingly, beyond business, PSI has also taken on a private commission from an individual investor (who may be a sci-fi fan).

The Fermi Explorer interstellar mission.

PSI announced: starting now, we are building a spacecraft!

According to the official announcement, they plan to launch the deep-space interstellar probe "Fermi Explorer" before 2029.

Using the optimal lightweight engineering solution optimized by the Emmy system within existing physical limits, this spacecraft will fly out of the solar system and, over 80,000 years, head toward the home of the Trisolarans and the Na'vi 4.2 light-years away, Centaurus.

Although perhaps no human will ever live to see that day, we wish them success.

Reference link: