with out a Hadamard matrix of order 668 and with a mathematical theorem I didn’t but belief sufficient to name a outcome.
Working with GPT-5.6 Sol, parallel brokers, exact-arithmetic packages and a proof assistant, I attacked two laborious issues over two days. One resisted each route I attempted. The opposite produced a proof candidate: a standard argument written out in full, precise checks on chosen inputs, a educating information and a partial formalisation in Lean.
That second end result remains to be a candidate. No specialist has reviewed the argument, its novelty is just not established, and Lean checks solely its algebraic core reasonably than the whole theorem. I’d not publish it as a brand new theorem in its present state.
The sample I took from the weekend is that mathematical experiments have gotten ample. The Hadamard mission closed 44 exactly outlined areas of the search area and audited 5 customary routes to a nonexistence proof. The Maxwell mission produced an entire proof candidate with a machine-checked algebraic core. Accepted mathematical information didn’t get cheaper alongside any of that, as a result of proof, understanding, novelty and impartial evaluate are separate obligations, and producing one other route settles none of them.
It is a concrete sequel to 2 earlier concepts. In AI Has Made Analysis Low cost. Understanding Is Nonetheless Costly, I argued that AI was making experimentation far cheaper than understanding. In From Tokens to Theorems: Constructing a Neuro-Symbolic AI Mathematician, I constructed a easy neuro-symbolic loop during which an LLM proposed mathematical formulation, SymPy checked them precisely, and failed candidates grew to become suggestions for the following try. That experiment was intentionally modest, nevertheless it uncovered an structure that reappeared right here at a a lot bigger scale.
Over this weekend, the identical sample prolonged past symbolic sequence issues. As an alternative of 1 mannequin and one symbolic checker, I used parallel brokers, exact-arithmetic packages, adversarial critics and a proof assistant. Candidate constructions and arguments have been generated, attacked, checked the place potential and both discarded or retained with an express standing. The loop had turn into richer as AI techniques improved via 2025 and 2026, however its primary construction remained the identical.
Wanting again afterwards, I realised the workflow intently resembled a broader sample that Jeff Dean just lately described for science and engineering. Fairly than viewing AI as a instrument for fixing one downside at a time, he characterised analysis itself as a recurring cycle: suggest an experiment, implement and run it, consider the result, then use that analysis to generate a greater experiment. His argument was that AI ought to automate this complete loop, decreasing iteration time, operating many experiments in parallel and studying from each analysis.
That was virtually precisely what had occurred over my weekend, albeit on a a lot smaller scale. The experiments have been mathematical reasonably than bodily. “Implementation” meant exact-arithmetic packages, constraint solvers or proof assistants as a substitute of laboratory tools. Analysis meant integer verification, counterexample search and partial formalisation reasonably than measuring a bodily system. Parallel brokers proposed constructions, generated verifiers, searched the literature, criticised arguments and translated concepts into completely different mathematical representations earlier than the following iteration started.
Seen this manner, essentially the most hanging improvement was not that AI produced a proof candidate or did not assemble a Hadamard matrix. It was that the experimental cycle itself had turn into partially automatable. As soon as precise evaluators existed, proposing, executing and refining mathematical experiments grew to become dramatically cheaper.
The essential distinction was not merely certainly one of scale however of validation. Within the earlier experiment, a candidate system might be examined immediately towards the sequence, though reproducing the noticed values didn’t essentially reveal the underlying arithmetic. Right here, verification grew to become layered. Precise arithmetic may reject incorrect constructions, Lean may certify elements of an argument, and literature searches may assist set up prior work, however none may alone decide novelty, validate each bridge in a common proof or resolve whether or not a outcome deserved to turn into accepted mathematical information.
Hadamard 668 resisted a precise search stack
The primary downside asks for a 668 by 668 grid crammed solely with +1 and -1, organized in order that any two completely different rows cancel out utterly: multiply them entry by entry, add up the outcome, and you should get precisely zero. In linear-algebra phrases, the rows are mutually orthogonal. It is a Hadamard matrix of order 668, and matrices of this type present up in error-correcting codes, sign processing and the design of experiments. In line with Epoch AI’s present catalogue of the issue, 668 is the smallest order for which no one is aware of whether or not one exists.
That is an interesting goal for computation as a result of a proposed reply is straightforward to verify. Multiply the candidate by its personal transpose utilizing precise integers, then learn off the outcome: each diagonal entry should be 668 and each different entry precisely zero. That could be a single matrix multiplication, and there’s no subjective judgement on the remaining gate.
Discovering one is one other matter. The grid has 446,224 entries and every is both +1 or -1, so a naive search has 2⁴⁴⁶,²²⁴ candidates to get via. No quantity of compute touches a quantity that measurement. All of the helpful work goes into not looking it, by imposing sufficient construction that the variety of free selections collapses.
That collapse is what the recognized building households purchase you. The primary route right here makes use of the truth that 668 is 4 occasions 167. Discover 4 quick +1/-1 sequences, of lengths 84 and 83, whose correlation errors cancel at each offset, and a normal recipe assembles them into the complete matrix. That recipe is carried out within the repository and was checked finish to finish on the smaller orders 28 and 36, the place precise integer checks affirm the ensuing matrices. This takes the search from 2⁴⁴⁶,²²⁴ candidates right down to 2³³⁴. Nonetheless far too many to enumerate, which is the trustworthy cause the issue is open, however now sufficiently small that construction and symmetry arguments can chew.
Different routes undergo completely different households, every with its personal identify: Legendre pairs of size 333, Williamson-type quadruples, cocyclic and transposed-Ito constructions over teams of order 167, and distinction households with prescribed symmetries. The names are protected to skim previous if they’re unfamiliar. What issues is that every is a special means of buying and selling one huge unstructured seek for a small structured one, and that every carried precise verifiers for the intermediate objects it accepted. Failed candidates have been rejected by integer arithmetic, not by approximate similarity.
A number of of the exclusions are exact sufficient to state numerically. A correlation certain proves that no answer lies inside Hamming distance 31 of 1 specific 334-bit start line: flipping any 31 or fewer of its bits can not attain a sound reply. The evaluate bundle will embody its full bit strings and digest. Individually, a precise enumeration covers all 4,096 quadruples constructed from circulant sequences, those the place every row is the earlier row rotated by a single place, restricted to these whose sample of destructive entries survives multiplying each index by 2 modulo 167. None of them works.
A counting argument over odd and even positions guidelines out another form: the case the place the third and fourth sequences are the primary two with the signal of each different entry flipped. Every of those exclusions covers solely the neighbourhood or symmetry sample it names, and nothing past it.
The mission additionally labored the issue from the opposite facet and tried to show that no such matrix exists on this household. That failed too, and the failure is particular sufficient to be helpful. Nonexistence outcomes on this space draw on a normal toolkit, and the audit labored via it merchandise by merchandise: Leung-Schmidt area descent, splendid factorisation and 2-adic valuation, Hasse-Minkowski, the multiplier theorems, and Bruck-Ryser-Chowla. Right here too the names matter lower than the result. Each one both fails to use at these parameters or applies and produces no obstruction. What stays is a brief listing of unresolved mathematical questions; none of those 5 customary routes provides the lacking obstruction.
Calibration mattered greater than any single run. The constraint solver was given a a lot smaller occasion of the identical four-sequence downside, one already recognized to have an answer: sequences of size 15 and 14, towards the goal’s 84 and 83. It didn’t discover that recognized answer inside two minutes. The timeout proved nothing in regards to the feasibility of both occasion. It confirmed that this solver setup couldn’t clear up even the smaller occasion throughout the two-minute calibration funds, so I ended treating longer runs of the identical setup as the principle route ahead.
By the tip, each closed area of the search area had been documented, along with an express listing of 9 remaining open questions. A number of promising heuristics proved ineffective. No Hadamard matrix of order 668 was discovered, and nothing right here advances the final conjecture. The issue stays open, and the eventual building might lie in a household this mission by no means explored. Even so, the journey was immensely rewarding. Exploring new arithmetic alongside an AI, making use of recognized methods to a longstanding open downside, gave a glimpse of upper arithmetic changing into more and more accessible and opening the door for extra folks to take part in mathematical discovery.
Maxwell’s downside produced a candidate that’s a lot tougher to verify
The second mission started with an previous downside from mathematical physics. Place some level expenses in area. Their mixed potential creates equilibrium factors the place the forces stability. The purpose is to certain what number of such factors there will be.
The final story modified just lately. A July 2026 paper by Arathoon, Ball and Kvalheim constructed 5 level expenses with no less than 24 nondegenerate equilibria, disproving Maxwell’s proposed normal system of (okay − 1)² for okay expenses. A second July paper by Gabrielov, Novikov, Novikov and Shapiro sharpened the certain for 3 constructive expenses from twelve to 6 nondegenerate equilibria, for each constructive Riesz exponent. Each papers can be found on arXiv: The Maxwell Conjecture is False and From 12 to six: Sharpening the Three-Cost Sure in Maxwell’s Drawback.
The precise candidate declare begins with three distinct supply factors in Euclidean area of dimension no less than two, three constructive expenses and a constructive exponent α. Their potential at a degree p away from the sources is:
The candidate declare is that this potential has at most 4 nondegenerate equilibrium factors for each α > 0. It treats collinear sources individually and claims precisely two equilibria there. The acquainted Coulomb potential is the case α = ½. That’s under the certain of six the July paper establishes, which is why the argument wants a specialist reader, not my very own confidence in it.
The geometric concept behind the candidate will be described with out the technical equipment. For 3 noncollinear expenses, each equilibrium lies inside their triangle. Its place will be represented by three constructive weights, known as barycentric coordinates. The candidate argument rewrites the bodily downside as a query in regards to the peaks of a associated mathematical floor.
If two of the related peaks existed, draw the straight chord becoming a member of them. Each endpoints must curve downwards alongside that chord. The candidate argument derives precise formulae for these two endpoint curvatures and makes use of a second inequality to indicate that they can’t each be destructive. If each bridge in that argument is sound, this leaves at most one nondegenerate native minimal of the bodily potential contained in the triangle. A planar index depend then provides the proposed higher certain of 4 equilibria.
The 2 initiatives require completely different validation. A single proposed Hadamard matrix will be settled by one precise calculation. The Maxwell candidate considerations each supply triangle, each set of constructive expenses and each constructive exponent, so it is determined by a series of quantified geometric, analytic and topological steps. A believable proof can disguise an error in any bridge between them.
The standard manuscript has been checked internally. Precise rational packages consider its derived identities on nonsymmetric rational inputs at α = ½ and α = 1. These are precise transcription controls on chosen examples, not a symbolic verification of the quantified identities.
The partial Lean improvement accomplished efficiently, and a supply scan discovered no sorry, admit or added axiom. Its 33 named theorems cowl the central second inequality, endpoint-gap algebra and an summary two-dimensional Hessian signal argument. They don’t but formalise the barycentric correspondence, the differential identities linking the bodily downside to the summary matrices, the endpoint-curvature derivation, the worldwide index and compactness arguments, or the collinear and higher-dimensional reductions. Lean is checking the algebra, and solely the algebra.
That boundary issues. Formalising the algebra that Lean accepts most readily can create an undeserved halo of confidence across the unformalised geometry. The subsequent formalisation goal ought to prioritise the steps most certainly to comprise an error.
The analysis loop grew to become partially automatable
I used the mannequin as one part in a analysis harness, alongside precise packages, a proof assistant and express proof guidelines. The preliminary exploration ran over a weekend, adopted by extra checking and writing. I chosen the issues, redirected or stopped unproductive routes, required the standing labels and determined which claims may seem right here. Recent-context critics obtained an artefact and an adversarial guidelines as a substitute of the entire working dialog. The deliberate public bundle will file the mannequin configuration, prompts, code variations and commit hashes wanted for a fuller provenance path.
One agent proposed formulations. One other tried to interrupt them. Others wrote precise verifiers, looked for counterexamples, translated bounded questions into constraint techniques, in contrast an argument with the literature, or defined an unfamiliar definition from a number of instructions. The work accrued via a repeatable loop:
- state a exact declare;
- derive penalties;
- check small or finite instances precisely;
- ask a fresh-context critic to assault the weak factors;
- formalise the parts the place machine checking provides confidence;
- replace the standing ledger;
- hold, revise or discard the concept.
Inside this specific harness, producing one other believable route was quick relative to checking it. Recording its precise scope, finding its weakest inference, checking whether or not it was already recognized and deciding whether or not it deserved knowledgeable consideration remained costly.
That loop may run throughout many branches in parallel. A failed method not consumed a whole night earlier than I learnt that its assumptions have been inconsistent. A conjectured id might be become a precise program and falsified shortly. A dense proof might be rewritten as a geometrical image, then as algebra, then as an inventory of obligations for formalisation. The Hadamard mission subsequently left a helpful file although it missed its goal: exact reductions, closed areas, failed methods and calibration outcomes that may stop the identical blind searches from being repeated.
This isn’t a conveyor belt working via a finite listing of open issues. Proving a theorem modifications the encompassing map. It exposes new constructions, suggests conjectures, connects questions that had seemed unrelated and creates new instructions to discover. Mathematical analysis is cyclical: a outcome closes one query whereas opening a number of others. Making every flip of that cycle cheaper can produce extra arithmetic to research, not an approaching finish to the topic.
The mannequin additionally accelerated studying
I got here to each initiatives with out specialist coaching of their underlying arithmetic. Over the weekend I bumped into aperiodic autocorrelation, algebraic norms, barycentric coordinates, Hessians, second inequalities and planar index principle.
The mannequin defined every concept repeatedly and from completely different angles. It moved between formulae, small numerical examples, visible instinct and the position an concept performed within the wider argument. When an evidence didn’t land, I may say so with out embarrassment and ask for an additional. Finally a number of items clicked into place.
Michael Nielsen describes a associated course of in Utilizing spaced repetition techniques to see via a chunk of arithmetic. His central level is that mathematical understanding is just not binary however will be deepened virtually indefinitely by decomposing a proof, restating its concepts in numerous kinds, probing variations and rebuilding the connections between them. With sufficient familiarity, the proof can start to really feel virtually clear: much less like a sequence of symbols to be adopted and extra like a set of mathematical objects whose relationships will be manipulated immediately.
My expertise with the mannequin felt like an interactive analogue of a part of that course of. As an alternative of setting up spaced-repetition playing cards, I may repeatedly ask for a similar concept to be represented as algebra, geometry, a numerical instance, an instinct or a problem query. The essential half was not that the mannequin provided an evidence as soon as, however that it made repeated reformulation low-cost. Nielsen calls the eventual state having the ability to “see via” a chunk of arithmetic. I didn’t attain something like specialist mastery over a weekend, however I may really feel the beginnings of the identical transition from following an argument line by line to recognising bigger chunks of its construction.
Earlier than the weekend, I realised, I had subconsciously considered mathematical analysis largely as fixing issues already ready on an inventory. The work itself felt completely different. A lot of it consisted of studying unfamiliar concepts, following failed routes far sufficient to grasp why they failed, reformulating questions and noticing surprising connections. A proof was one milestone in that bigger exploratory course of, not the method as a complete.
The result was working literacy reasonably than experience. I may comply with the construction effectively sufficient to ask higher questions, discover when two claims had been conflated and perceive what an exterior reviewer would want to examine. A transparent rationalization nonetheless offers no proof {that a} theorem is true, and I needed to hold reminding myself of the distinction between following an argument and having the ability to reconstruct it.
For me, this was essentially the most rapid type of mathematical abundance. The mannequin met me on the level the place understanding stopped and stored altering the illustration till progress resumed. It made an unfamiliar topic explorable with out making me a specialist in it.
In Mathematicians are grappling with the likelihood that AI would possibly eclipse them for Understanding AI, Kai Williams stories on interviews with greater than twenty mathematicians. What he heard most frequently was not fashions proving issues, however fashions used to discover a means into unfamiliar areas of the literature. That matches my expertise precisely. The mannequin provided a map and several other translations; the underlying papers, derivations and precise checks nonetheless needed to carry the declare.
Tasmin Chu’s essay Mathematicians have to act identifies a danger in the identical workflow. She argues that the modest extensions, literature workouts and first proofs most uncovered to automation are additionally how college students turn into researchers. If a mannequin does that work as a substitute of directing it, the sector might weaken that pipeline. My very own extension of her level is that additionally it is the place future reviewers be taught judgement. One weekend can not settle the priority. I learnt most after I needed to reconstruct an argument, ask what would falsify it and uncover why an evidence had did not land, not after I merely obtained a solution.
Bigger initiatives present the identical uneven automation
Two a lot larger efforts level the identical means.
In 2025, Google DeepMind reported that AlphaEvolve had been utilized to greater than 50 open mathematical issues. In line with DeepMind, it rediscovered one of the best recognized answer in roughly three quarters of the instances and improved one of the best recognized lead to a couple of fifth. The strategy applies to a helpful however slender class of issues whose proposed options will be expressed as algorithms and scored routinely.
In Might 2026, OpenAI reported {that a} general-purpose mannequin discovered a counterexample to the Erdős unit distance conjecture. 9 exterior mathematicians then produced a brief, digested, human-verified model of the argument, which put the human verification into the revealed file reasonably than leaving it as a declare in regards to the mannequin.
Each examples present that automation is uneven. Machine-readable scoring lets a system reject unhealthy candidates and iterate with out ready for an individual to learn each. A common proof candidate nonetheless requires cautious work on ideas, translations and quantified arguments.
Tom Zahavy’s ICML 2026 place paper LLMs can’t bounce provides a vocabulary for that cut up. Borrowing Peirce’s three modes of inference, he argues that machine studying has mechanised induction, which finds a rule by compressing many examples, and is quickly mechanising deduction, which derives penalties from premises already mounted. What it has not mechanised is abduction: proposing the premise itself to clarify one thing stunning. His case research is normal relativity, and his central commentary is that Newtonian gravity confronted no measurable disaster on the time. The equivalence of inertial and gravitational mass had been checked to a precision of10⁻⁹, and the one anomaly, Mercury’s orbit, was broadly attributed to an undiscovered planet. An optimiser would have discovered virtually nothing to enhance. He grants {that a} mannequin given Einstein’s 1915 postulates may plausibly derive the sector equations, since that half is deduction; the 1913 model failed as a result of the axioms have been flawed, not as a result of the logic was.
The paper is a place piece and says so, and Zahavy is express that his argument is aimed on the bodily sciences, the place the uncooked materials is sensory, and that arithmetic grounds its intuitions otherwise. I’d not push the analogy previous that. However the division of labour was recognisable from the within. The precise packages and Lean did deductive work effectively, and the Maxwell candidate progressed as a result of a framing already existed to infer inside. Hadamard 668 had no such framing handy, and what it lacks is just not extra search however a premise: a building household or a theorem no one has written down. His studying of AlphaEvolve makes the identical level from the opposite facet: it optimises effectively inside a hard and fast framework as a result of it has a gradient to comply with.
The tempo has not slowed since. Williams opens his article with a Fields Medallist becoming a member of OpenAI’s security staff and an inner mannequin that the corporate says solved ten main open issues; Chu opens hers with the identical announcement, noting that the reported token price of these outcomes was round two thousand {dollars}. I’ve not checked both declare, and neither had been via evaluate after I wrote this.
Jordan Ellenberg captured the optimistic historic response in his 2014 e-book How To not Be Improper, in a passage Williams quotes: “We’ll reclassify that analysis as ‘computation.’” Arithmetic has repeatedly moved its frontier after machines absorbed an previous process. Reclassification doesn’t, by itself, clarify how folks will be taught to decide on the following questions, choose the solutions or maintain a neighborhood round them.
AI might subsequently enhance mathematical abundance in a broader sense than proving extra theorems. It could generate extra conjectures, partial theories, proposed connections between fields and instructions price investigating. My weekend doesn’t set up how typically these strategies will likely be deep or genuinely new; it exhibits solely how cheaply candidate routes can already be generated and examined. Even a modest enhance would shift extra of the burden towards understanding, curation, evaluate, rationalization and prioritisation. The problem is just not solely to provide extra arithmetic, however to resolve what deserves consideration.
Jeff Dean, in founding Discoveryloop, frames this as a programme extending far past arithmetic. Fusion, medication, cybersecurity and supplies science all share a model of the identical experimental cycle: suggest an experiment, run it, consider the outcome and use that analysis to decide on the following experiment. If AI more and more automates that loop throughout science and engineering, arithmetic could also be an early illustration of a broader transition reasonably than a particular case. The bottleneck then strikes from producing experiments to deciding which outputs are dependable, vital and worthy of changing into trusted information.
Abundance creates a human and institutional downside
If fashions can generate many extra conjectures, proof candidates, counterexamples and partial formal developments than people can learn, storing them in chat transcripts is not going to work. The identical acceleration additionally modifications who learns the craft, who receives credit score, who takes duty for errors and who’s requested to evaluate the output.
A helpful mathematical file wants greater than a title and a PDF. It ought to carry a normalised assertion, express assumptions, a standing label and the exact scope of any computation. It also needs to carry the formal artefacts the place they exist, the dependencies on earlier outcomes, the provenance of the human and machine contributions, the novelty standing, the recognized assaults and a plain-language rationalization.
Search ought to function over claims and dependencies reasonably than solely key phrases. A helpful system may match a proposed lemma towards equal statements in numerous notation, present which unreviewed claims would suggest a goal outcome, and determine arguments that depend on the identical unformalised bridge. Failed routes needs to be searchable too, supplied their scope is precise. Over time, this might turn into a shared corpus of profitable proofs, failed makes an attempt, deserted search branches, reductions, counterexamples, calibration outcomes and intermediate constructions. Each well-scoped failure may slender the remaining search area as a substitute of being unknowingly repeated.
With out that infrastructure, I anticipate repeated work and misplaced confidence. Fashions will rediscover previous outcomes, subtly fluctuate false proofs and generate extra materials than peer evaluate can take in. The bottleneck would transfer from producing arithmetic to curating a reliable map of it.
Williams’s interviews have been reporting, not a consultant survey, and the folks he spoke with provided no single response. Many anticipated AI to enhance their work within the close to time period and already used it in restricted methods, particularly to traverse unfamiliar literature. Others fearful in regards to the coaching path for youthful researchers, future funding and a occupation organised round duties that fashions might be taught to carry out. Their disagreement displays a number of objectives being bundled collectively. Fixing open issues is one intention of arithmetic, however so are understanding, rationalization, theory-building, educating and sustaining a neighborhood.
The Leiden Declaration warns that present automated methods produce believable however unreliable arguments which are laborious to tell apart from appropriate proofs, that the identical downside applies to formalisations via the interpretation between machine and human shows, and that AI-assisted papers make reviewing extra demanding. Timothy Gowers, writing about the declaration, pushes the thought additional: he imagines mathematicians deciding on from an enormous physique of AI-generated arithmetic and writing it up in order that others can take in it, and he’s candid about how a lot of the present tradition that will displace.
Any such map will want status and incentives behind it. Gowers makes the purpose sharply: if one individual will get a mannequin to resolve an open downside and one other digests the answer and explains it in order that mathematicians can be taught from it, the second individual ought to get a lot of the credit score. Explaining why a generated outcome issues, discovering the delicate flaw in it, or connecting it to a uncared for theorem could also be price greater than producing the primary draft. At present’s publication tradition is just not designed to recognise these contributions cleanly.
The attribution downside reaches additional again than the ultimate proof. Many celebrated breakthroughs are the fruits of many years of definitions, lemmas, conjectures, failed approaches and methods developed by many researchers. If an AI provides the final lacking argument, rewarding solely that final mile dangers obscuring the human work that made the step potential and gave it that means. Provenance ought to subsequently file not solely who or what accomplished the proof, however the chain of concepts and outcomes on which it relies upon. Credit score ought to comply with that chain reasonably than accumulate at its endpoint.
Chu’s essay sharpens all of this into an institutional argument. She presents correct attribution, duty for errors and open information as mathematical values that AI use might pressure. A generated proof can obscure the human work it is determined by, and a believable error can eat way more knowledgeable consideration than it took to provide. She additionally worries that researchers will share fewer unfinished concepts if someone else can convert these concepts into model-assisted outcomes earlier than their originators end. That final level is a forecast reasonably than one thing my experiment demonstrates, nevertheless it means a declare registry would want guidelines for consent and attribution round unpublished concepts, not solely higher search over completed ones. Among the many prescriptions she attracts from which are that mathematicians ought to use fashions judiciously or under no circumstances, and that chat logs needs to be revealed alongside mathematical papers.
Impartial evaluate can be the following pure step
I don’t intend to show this private weekend studying mission right into a public evaluate effort. If I did, I’d first set up the literature and precedence place, invite specialists to assault the traditional argument, and formalise the steps most inclined to undetected error. Till these gates are handed, it stays a proof candidate.
For Hadamard 668, a remaining precise verifier has been written however has not but been exercised on a real candidate. The lacking ingredient is a brand new structural building or theorem. The destructive file ought to information future makes an attempt with out being mistaken for an answer or a nonexistence proof.
Whether or not or not both mission goes additional, the sturdy object is just not a proof alone however its place on the map. A proof sitting on a server has surprisingly little worth if no one is aware of it exists, why it issues, the way it pertains to earlier work or whether or not it may be trusted. As mathematical artefacts multiply, navigation via them turns into the scarce useful resource: finding the related declare, understanding its dependencies, judging its standing and explaining the place it modifications the frontier.
Working the loop quicker doesn’t shut that hole. Low cost experiments multiply the issues we are able to verify and the questions we are able to ask; they don’t, on their very own, inform us which premises are price checking or flip an output into shared understanding. Experiments turn into low-cost. Information doesn’t.
Disclaimer: This was a casual, curiosity-driven mission undertaken in my private time. The views expressed listed below are my very own and don’t characterize these of my employer or any affiliated organisation. The content material relies on private expertise and reflection, and shouldn’t be taken as skilled or tutorial recommendation.
📚References
- Gabrielov, A., Novikov, Dm., Novikov, T., & Shapiro, B. (2026). From 12 to six: Sharpening the Three-Cost Sure in Maxwell’s Drawback. Proves
that three constructive level expenses admit at most six nondegenerate equilibria for each constructive Riesz exponent, enhancing the twelve-point certain from the authors’ earlier work by supplying a separation argument on the saddle of a separated-variable first integral. That is the revealed certain the proof candidate mentioned right here would sit under, and the explanation that candidate wants a specialist reader reasonably than my very own confidence in it. - Arathoon, P., Ball, G., & Kvalheim, M. D. (2026). The Maxwell Conjecture is False. Reveals 5 level expenses whose electrostatic potential has no less than 24 important factors, all nondegenerate, refuting Maxwell’s proposed certain of (okay−1)² for okay expenses. Along with the paper above it defines the state of Maxwell’s downside in July 2026, which is the setting for the four-equilibrium declare described right here.
- Zahavy, T. (2026). Place: LLMs can’t bounce. ICML 2026, PMLR 306. Argues through Peirce’s three modes of inference that machine studying has mechanised induction and is quickly mechanising deduction, however not abduction — the invention of the premise itself — utilizing normal relativity as a case research during which no error sign existed for an optimiser to comply with. Provides the vocabulary this text makes use of to separate the deductive work Lean did effectively from the lacking premise that retains Hadamard 668 open.
- Chu, T. (2026, 2 August). Mathematicians have to act. Units out attribution, accountability for error and open information as neighborhood values that AI use might pressure, argues that the tractable issues most uncovered to automation are additionally how analysis mathematicians are skilled, and takes the place that within the present local weather it’s unethical to ask a mannequin to show a brand new theorem. The usual towards which this mission is measured within the closing part.
- Williams, Ok. (2026, 4 August). Mathematicians are grappling with the likelihood that AI would possibly eclipse them.Understanding AI. Reporting from the Worldwide Congress of Mathematicians based mostly on conversations with over twenty mathematicians, discovering the commonest use to be navigating unfamiliar areas of the literature reasonably than proving outcomes, and quoting the passage from Ellenberg’s How To not Be Improper reproduced right here.
- Nielsen, M. A. (2019, 12 January). Utilizing spaced repetition techniques to see via a chunk of arithmetic. Cognitive Medium. Describes mathematical understanding as an open-ended means of repeatedly decomposing, reformulating and connecting the weather of a proof till they turn into deeply internalised. Nielsen makes use of spaced repetition because the mechanism, arguing that the method can ultimately produce a way of having the ability to “see via” the arithmetic reasonably than merely reproduce its symbolic steps. This offers a helpful analogue for the repeated, multi-representation explanations used through the studying course of described right here.
