Alex Wissner-Gross, a co-founder of the AI physics start-up Physical Superintelligence, published an article on X on 24 October 2026 announcing Chain Theory, which Wissner-Gross calls “a new research program of mine that seeks to rebuild fundamental physics from quantum information”. The article was read on 24 October 2026, with the working paper it links, dated 23 October 2026.
The article announces as its first result “a parameter-free candidate for the fine-structure constant, derived from a single axiom, that matches its measured value to ten significant digits.” The paper’s acknowledgments say the work “was developed as part of a long-term human-superintelligence collaboration supervised by the human author.”
Business Pill · THE AI AUDIT PROBLEM
A short Business Engineer explainer on why checking AI-made work can cost more than making it. It teaches the general idea only and says nothing about Chain Theory.
The key insight: As we read it, the paper makes two kinds of claim: an arithmetic one, that a stated formula yields a number matching CODATA’s value to ten significant digits, which can be checked today, and a physical one, that the number is the fine-structure constant, which Wissner-Gross calls a conjecture.
The fine-structure constant sets the strength of electromagnetism. The Standard Model of particle physics takes its value from experiment rather than deriving it; Wissner-Gross writes that it is “one of the most precisely measured numbers in all of science”.
The candidate is the algebraic number (79 + 48√2 + 42√3 + 3√6)/31104, which Wissner-Gross describes as “a weighted count over the 172 measurement records of a single entangled pair, small enough to check by hand.” The paper gives its inverse as 137.035 999 190.
Item
Value
Source
Chain Theory candidate, inverse
137.035 999 190
Wissner-Gross paper, dated 23 Oct 2026
CODATA 2022 recommended inverse
137.035 999 177, standard deviation 21 in the last two digits
As cited in the X article
Distance from CODATA
0.60 standard deviations
X article; our arithmetic agrees
Matching digits
First ten significant digits
X article
Our arithmetic: evaluating the stated formula gives an inverse of 137.0359991896, which is 0.60 standard deviations from 137.035 999 177 with a standard deviation of 0.000 000 021. This checks the arithmetic only, not the physics.
Wissner-Gross writes that the number is “checked in Lean rather than fitted, and only its identification with α(0) is conjectured.” Lean is a proof assistant, software that verifies each step of a mathematical proof.
The inverse fine-structure constant: CODATA 2022’s recommended value, 137.035 999 177, against the Chain Theory candidate, 137.035 999 190, which Alex Wissner-Gross’s paper of 23 October 2026 puts 0.6 standard deviations away. Bars show the shared digits; the link between the candidate and the constant is the author’s conjecture.
Alex Wissner-Gross shared the Chain Theory article on X on 24 October 2026, announcing a research program to rebuild fundamental physics from quantum information and, as its first result, a parameter-free candidate for the fine-structure constant that the article says matches the measured value to ten significant digits.
Where AI Comes In
The X article does not describe how the work was done. The paper’s acknowledgments do: “This work was developed as part of a long-term human-superintelligence collaboration supervised by the human author. It was assisted by resources of Physical Superintelligence PBC, in which the author has a financial interest.” The same wording closes the 754-page Chain Theory compendium linked from the article.
Neither text names an AI model, says which parts an AI system produced or says how much compute was used. “Superintelligence” is the author’s term.
Unite.AI reported on 1 September 2026 that Matt Pines, Alex Klokus and Wissner-Gross launched Physical Superintelligence with $58 million in seed funding led by Breakthrough Energy Ventures. It reported that the Cambridge, Massachusetts company described itself as an AI-native physics lab staffed by virtual physicists. “PSI aims to industrialize the discovery of new physics,” chief executive Matt Pines said, according to Unite.AI.
How It Was Checked
Wissner-Gross writes that experiments will settle it. The measurements of the constant “still disagree with one another by more than their stated uncertainties, so this comparison is not yet decisive.” The article says new measurements at Northwestern University, Berkeley and Paris “aim to be up to ten times more precise” and, if they agree, “will either close in on the derived value or rule it out.”
The article also stops short of a full claim: “I do not yet claim to have done the same,” Wissner-Gross writes of rebuilding physics, adding that “Chain Theory is new, much of it is still conditional or underdeveloped”.
Connecting the Coverage
Today, our Terence Tao piece reported Tao’s view, from an interview with The Observer quoted on Tao’s own page, that big prizes and goals in mathematics have become benchmarks for AI labs.
Our OpenAI manuscripts piece reported on 6 October that OpenAI’s internal model produced 722 mathematics manuscripts, later 719, and that OpenAI’s own history file put the share of top-line results formalized at about 42%. Formal checking in Lean is the same tool Wissner-Gross cites.
On 15 October, our Sakana AI Scientist piece reported Sakana AI’s The Agentic Scientist, a system that runs computational research with coding agents; in one test, a Kimi K3 model on the Pi harness reached scores similar to Opus 4.7 on Claude Code while cutting API costs by 75%, the paper says.
What is new here: a physicist publishing a candidate value for a constant of nature, crediting a human-superintelligence collaboration and an AI physics company in which the author has a financial interest, with the arithmetic checked in Lean and the physical claim left as a conjecture.
The Structural Read
As we read it, the AI angle rests on the acknowledgments: the paper credits a long-term human-superintelligence collaboration supervised by the human author, and the resources of Physical Superintelligence, without naming a model or sizing its part.
As we read it, the Lean check covers the computation of the number, as the paper’s abstract says, not whether the number is the constant; Wissner-Gross writes that new measurements will either close in on the value or rule it out.
As we read it, the author is also a founder: Unite.AI reported that Wissner-Gross co-founded Physical Superintelligence, which raised $58 million in seed funding, and the paper discloses a financial interest.
Alex Wissner-Gross, ‘Chain Theory’, X Article, 24 October 2026
“If they agree with one another, they will either close in on the derived value or rule it out.”
Three Implications
FOR RESEARCH LEADERS TRACKING AI FOR SCIENCE The paper credits a human-superintelligence collaboration but gives no model, workflow or compute figures, so the AI share of the work cannot be measured from it.
FOR INVESTORS IN AI SCIENCE START-UPS Unite.AI reported that Physical Superintelligence raised $58 million in seed funding led by Breakthrough Energy Ventures; the paper discloses the author’s financial interest in the company.
FOR ANYONE CITING THE RESULT Wissner-Gross writes that the number’s identification with the fine-structure constant is conjectured and that new measurements could rule it out.
The framework puts it as “Agentic use is a demand shock, not a usage tweak.” Its examples include research output inside OpenAI rising 56 times in seven months.
As we read it, a researcher crediting a human-superintelligence collaboration for a 754-page compendium points the same way, though neither the paper nor the framework measures the AI’s share. The framework was published in July 2026 and does not discuss this work.
What Is Not Established
We read the X article, the paper’s abstract, introduction and acknowledgments, and the compendium’s acknowledgments. Neither the article nor the paper says the work has been peer reviewed, and we did not find a journal publication. We did not check the physics, the 172-record count or the Lean proof; our arithmetic checks only that the stated formula gives the stated number.
The texts do not name the AI system or measure its contribution. Wissner-Gross has a financial interest in Physical Superintelligence, as the paper discloses. We did not contact Wissner-Gross or the company.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Alex Wissner-Gross says a single quantum-information axiom yields a number that matches the measured fine-structure constant to ten significant digits, with the arithmetic checked in Lean and the link to the constant left as a conjecture. The paper credits a human-superintelligence collaboration and Physical Superintelligence, the AI physics company Wissner-Gross co-founded.
A note on sourcing. This piece rests on Alex Wissner-Gross’s X Article of 24 October 2026, read in full, and the paper and compendium it links on alexwg.org, whose abstract, introduction and acknowledgments we read. Company details come from Unite.AI. We did not check the physics or contact Wissner-Gross or Physical Superintelligence. Earlier figures come from our own reporting, linked in the text.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
