Background

Large-scale semantic processing and strong computer assistance of mathematics and science is our inevitable future. New combinations of AI and reasoning methods and tools deployed over large mathematical and scientific corpora will be instrumental to this task. The AITP conference is the forum for discussing how to get there as soon as possible, and the force driving the progress towards that.

Topics

Sessions

There will be several focused sessions on AI for ATP, ITP, mathematics, relations to general AI (AGI), Formal Abstracts, linguistic processing of mathematics/science, modern AI and big-data methods, and several sessions with contributed talks. The focused sessions will be based on invited talks and discussion oriented. AITP'26 is planned as an in-person conference.

Confirmed Participants/Speakers (TBC)

João AraújoUniversidade Nova de Lisboa
Michael R. DouglasStony Brook University
Mario CarneiroChalmers University and University of Gothenburg
Simon FriederUniversity of Oxford
Thibault GauthierAI4REASON
Ben GoertzelSingularityNET
Georges GonthierINRIA
Sean HoldenUniversity of Cambridge
Jan JakubuvCzech Technical University in Prague
Mikoláš JanotaCzech Technical University in Prague
Moa JohanssonChalmers University and University of Gothenburg
Cezary KaliszykUniversity of Melbourne
Peter KoepkeUniversity of Bonn
Konstantin KorovinThe University of Manchester
Michael KinyonUniversity of Denver
Miroslav OlsakUniversity of Cambridge
Jan von PlatoUniversity of Helsinki
Auguste PoirouxEPFL and Math Inc
Aarne RantaChalmers University and University of Gothenburg
Michael RawsonUniversity of Southampton, UK
Stephan SchulzDHBW Stuttgart
David StanovskýCharles University
Martin SudaCzech Technical University in Prague
Christian SzegedyAletheAI
Josef UrbanAI4REASON and University of Gothenburg
Robert VeroffUniversity of New Mexico
Andrei VoronkovEasychair and University of Manchester
Zsolt ZomboriAlfréd Rényi Institute of Mathematics

Invited talks (TBC)

Simon FriederOpen Weights Are Not Enough: Reproducing an IMO26 Medal with FM-Pochi-32B
Ben GoertzelFrom AITP to AGITP: Toward a Neural-Symbolic Architecture for Creative Math Conjecturing
Michael KinyonAutomated Deduction in Quasigroup and Loop Theory Or: How I Learned to Stop Worrying and Start Letting Computers Prove Theorems
Jan von PlatoDeciphering Goedel
Auguste Poiroux(Auto-)Formalization of the Fields Medal Sphere Packing Results (TBC)
Christian SzegedyTBA
Robert VeroffAutomating the search for proofs of open conjectures
Andrei VoronkovTBA

Dates

AITP solicits contributed talks. Selection of those will be based on extended abstracts/short papers of 2 pages (excluding references) formatted with easychair.cls. Submission is via EasyChair. Accepted contributions will be published in an informal book of abstracts available from this website. The extended abstracts are considered non-archival. The contributed talks have to be presented in-person.

Contributed talks

Shane Caldwell. ProofJudge: Tool-Grounded LLM Evaluation of Formal Proof Quality in Mathlib
Nil Geisweiller. Using Forward Chaining to Go Backward
Romana Jezek. OConcise – Towards an automatic mathematical assistant
Autumn Mapes. Visual Lean: An accessible interface for Lean 4 proofwriting
Peter Koepke and Joshua Wirtz. Naproche Natural Language Formalizations with LLMs
Silvère Gangloff, Jan Hula and Mirek Olsak. Generating Lean 4 Tactics from User Specifications: Strategies for LLM-Assisted Metaprogramming
Karel Chvalovský, Martin Suda and Josef Urban. Teaching Vampire New Tricks: An Experimental Study of Neural Clause Selection
Jan Jakubuv, Cezary Kaliszyk and Martin Suda. Learning-Guided Higher-Order Automated Reasoning for Isabelle/HOL
Christoph Wernhard. Generating Theorems by Generating Proof Structures
Keneni W. Tesema, Jan Jakubuv and Martin Suda. Premise Selection for Higher-Order ATP Problems
Christoph Wernhard and Zsolt Zombori. Are Proof Structures Machine-Learnable? A Study with DAG-Structured Proof Terms
Andrew Fish, Neslihan Gugumcu and Alexei Lisitsa. Towards a Theory of Affine and Polynomial Biquandles: Enumeration, Ring-Dependence, and Beyond
Levente Rácz. Mitigating State Dilution in Axiomatic Proof Trees via Block Attention Residuals
Thibault Gauthier. Program Generation as an Evolution Operator
Guy Axelrod. Learning Abstractions for Guiding Equational Proof Search
Weichen Winston Yin. Quiver: A Knowledge-Graph Extraction Pipeline for Lean Libraries
Aarne Ranta and Jan von Plato. From Proof Objects to Linear Proof Texts
Aarne Ranta and Adrian De Lon. Reinformalizing Autoformalizations: Combining Naproche and Informath
J.D. Phillips. Equation wrangling: How I learned to stop worrying about large equational sets and love automated deduction
Atle Hahn and Adrian De Lon. Autoformalization Experiments with Felix
Henri Nikoleit, Ankit Anand, Anurag Murty Naredla and Heiko Röglin. The Art of Being Difficult: Combining Human and AI Strengths to Find Adversarial Instances for Heuristics
Michael Friedman. How Opaque are AI-generated Proofs? Past and Contemporary Anxious Reactions to the Mechanization of Mathematics
Cameron Freer. From Prompts to Protocols: lean4-skills for AI-Assisted Lean Formalization
Alessandro Sosso, Akhil Arora and Bas Spitters. Agentic Proving for Program Verification
Michael Beeson. Autoformalizing Euclid
Adam Dingle. Towards Automatic Verification of Textbook Proof Steps
Matěj Kripner and Milan Straka. NanoProof: Open and Efficient Theorem Proving in Lean 4
Miroslav Olšák and David Stanovsky. Project Proposal: Automated Correction of Math Assignements
Guillaume Baudart and Assia Mahboubi. A Case Study in Mechanizing Computer-Aided Proofs: Small Gaps Between Primes
Masaya Taniguchi and Sho Sonoda. Neural-Guided Kripke Countermodel Synthesis with Symbolic Verification
Ilija Ivanov, Jiayi Wu, Gavin Zhao, Zekai Li, Stephen Bach and Robert Lewis. Mining Machine-Generated Lean Proofs: Advice for Users, Developers, and Testers
Luis Berlioz and Paul-André Melliès. Vector Space Representations of Equational Theories
Mario Carneiro, Steven D. Miller, Magnus Myreen and Josef Urban. Autoformalizing Fast–Slow Equivalence Proofs for Atlas Computations in the Unitary Dual Problem
Dustin Bryant, Jonathan Julian Huerta Y Munive, Cezary Kaliszyk and Josef Urban. Munkres’ General Topology Autoformalized in Isabelle/HOL
Zarathustra Goertzel and Oruži Ai. Lean 4 Verified Metamath Verifier: Investigating AI-driven Software Verification
Jan Jakubuv, Miroslav Olšák, Martin Suda and Josef Urban. Neural Conjecturing for Saturation Theorem Provers
Mario Carneiro. AI-assisted formalization of Pi injectivity and uniqueness of typing for Lean's kernel
András Z. Salamon and Michael Wehar. Long, wide, and trivial: a nondeterministic time hierarchy in Isabelle (progress report)
Freek Wiedijk. Autoformalization using Formal Proof Sketches
Jan Hůla, Mirek Olsak and Petr Hyner. Making the Implicit Explicit

Program Committee (TBC)

Mantas BakšysUniversity of Cambridge
Guillaume BaudartINRIA
Jasmin Christian BlanchetteLMU Munich
David CernaDynatrace
Michael R. Douglas (co-chair)Stony Brook University
Ulrich FurbachUniversity of Koblenz
Thibault GauthierAI4REASON
Aishik GhoshGeorgia Tech
Georges GonthierINRIA
Zarathustra GoertzelCzech Technical University in Prague
Thomas C. Hales (co-chair)University of Pittsburgh
Sean HoldenUniversity of Cambridge
Mikoláš JanotaCzech Technical University in Prague
Moa JohanssonChalmers University and University of Gothenburg
Cezary Kaliszyk (co-chair)University of Melbourne
Michael KinyonUniversity of Denver
Peter KoepkeUniversity of Bonn
Konstantin KorovinThe University of Manchester
Mirek OlsakCharles University
Jelle PiepenbrockEindhoven University of Technology
Bartosz PiotrowskiIDEAS NCBR
Michael Rawson (co-chair)University of Southampton, UK
Stephan Schulz (co-chair)DHBW Stuttgart
Sho SonodaRIKEN AIP
Martin SudaCzech Technical University in Prague
Josef UrbanAI4REASON and University of Gothenburg
Zsolt ZomboriAlfréd Rényi Institute of Mathematics


Organizers

Georges GonthierINRIA
Cezary KaliszykUniversity of Melbourne
Josef UrbanCzech Technical University in Prague


AITP Program



August 30
19:30 dinner


August 31
9:00-10:15 Chair: Josef Urban Welcome

Auguste Poiroux
(Auto-)Formalization of the Fields Medal Sphere Packing Results (TBC) (50m)

Cameron Freer
From Prompts to Protocols: lean4-skills for AI-Assisted Lean Formalization (20m)

10:15-10:45 coffee break
10:45-12:05 Chair: Peter Koepke Jan von Plato
Deciphering Goedel (45m)

Aarne Ranta and Jan von Plato
From Proof Objects to Linear Proof Texts (20m)

Autumn Mapes
Visual Lean: An accessible interface for Lean 4 proofwriting (15m)

12:05-13:30 lunch
16:30-17:00 coffee break
17:00-19:00 Chair: Georges Gonthier Guillaume Baudart and Assia Mahboubi
A Case Study in Mechanizing Computer-Aided Proofs: Small Gaps Between Primes (20m)

Alessandro Sosso, Akhil Arora and Bas Spitters
Agentic Proving for Program Verification (25m)

Weichen Winston Yin
Quiver: A Knowledge-Graph Extraction Pipeline for Lean Libraries (25m)

Mario Carneiro
AI-assisted formalization of Pi injectivity and uniqueness of typing for Lean's kernel (25m)

remote speakers?
remote talks (25m)

19:00 dinner


September 1
9:00-10:15 Chair: Stephan Schulz Andrei Voronkov
TBA (50m)

Karel Chvalovský, Martin Suda and Josef Urban
Teaching Vampire New Tricks: An Experimental Study of Neural Clause Selection (25m)

10:15-10:45 coffee break
10:45-12:05 Chair: Andrei Voronkov Jan Jakubuv, Cezary Kaliszyk and Martin Suda
Learning-Guided Higher-Order Automated Reasoning for Isabelle/HOL (25m)

Keneni W. Tesema, Jan Jakubuv and Martin Suda
Premise Selection for Higher-Order ATP Problems (25m)

Guy Axelrod
Learning Abstractions for Guiding Equational Proof Search (25m)

12:05-13:30 lunch
16:30-17:00 coffee break
17:00-19:00 Chair: Michael Rawson Michael Kinyon
Automated Deduction in Quasigroup and Loop Theory Or: How I Learned to Stop Worrying and Start Letting Computers Prove Theorems (45m)

J.D. Phillips
Equation wrangling: How I learned to stop worrying about large equational sets and love automated deduction (10m)

Robert Veroff
Automating the search for proofs of open conjectures (45m)

Luis Berlioz and Paul-André Melliès
Vector Space Representations of Equational Theories (20m)

19:00 dinner


September 2
9:00-10:15 Chair: Thibault Gauthier Freek Wiedijk
Autoformalization using Formal Proof Sketches (30m)

Adam Dingle
Towards Automatic Verification of Textbook Proof Steps (20m)

András Z. Salamon and Michael Wehar
Long, wide, and trivial: a nondeterministic time hierarchy in Isabelle (progress report) (20m)

10:15-10:45 coffee break
10:45-12:25 Chair: Cameron Freer Ben Goertzel
From AITP to AGITP: Toward a Neural-Symbolic Architecture for Creative Math Conjecturing (45m)

Thibault Gauthier
Program Generation as an Evolution Operator (15m)

Zarathustra Goertzel and Oruži Ai
Lean 4 Verified Metamath Verifier: Investigating AI-driven Software Verification (20m)

Shane Caldwell
ProofJudge: Tool-Grounded LLM Evaluation of Formal Proof Quality in Mathlib (20m)

12:30-14:00 lunch
17:00-17:30 coffee break
17:30-19:00 Chair: TBA Select converts give their testimony (passionate testimonies welcome!)
Praising Our New AI Lord(s) Or: How Chatgpt/Claude/Codex Delivered Me From Cognitive Labor and Helped Me Solve Impressive Math/Other Problems (30m) (advertisment here

Panelists TBC
Panel discussion: Future of Math/CS/Science (and their current education/publishing/other systems/institutions) in the age of AI (60m)

Submit your names for the testimony and submit panel questions here)

19:00 dinner


September 3
9:00-10:15 Chair: Mirek Olšák Simon Frieder
Open Weights Are Not Enough: Reproducing an IMO26 Medal with FM-Pochi-32B (50m)

Michael Friedman
How Opaque are AI-generated Proofs? Past and Contemporary Anxious Reactions to the Mechanization of Mathematics (25m)

10:15-10:45 coffee break
10:45-12:05 Chair: Freek Wiedijk Peter Koepke and Joshua Wirtz
Naproche Natural Language Formalizations with LLMs (25m)

Adrian De Lon
Felix: a Scalable Natural Proof Assistant (15m)

Atle Hahn and Adrian De Lon
Autoformalization Experiments with Felix (20m)

Aarne Ranta and Adrian De Lon
Reinformalizing Autoformalizations: Combining Naproche and Informath (15m)

12:10-13:30 lunch
16:30-17:00 coffee break
17:00-19:00 Chair: Martin Suda Christoph Wernhard
Generating Theorems by Generating Proof Structures (25m)

Christoph Wernhard and Zsolt Zombori
Are Proof Structures Machine-Learnable? A Study with DAG-Structured Proof Terms (25m)

Andrew Fish, Neslihan Gugumcu and Alexei Lisitsa
Towards a Theory of Affine and Polynomial Biquandles: Enumeration, Ring-Dependence, and Beyond (25m)

Masaya Taniguchi and Sho Sonoda
Neural-Guided Kripke Countermodel Synthesis with Symbolic Verification (25m)

Mario Carneiro
Thinking Sand: A plan to verify a computer down to physics (20m)

19:00 dinner


September 4
9:00-10:15 Chair: Aarne Ranta Henri Nikoleit, Ankit Anand, Anurag Murty Naredla and Heiko Röglin
The Art of Being Difficult: Combining Human and AI Strengths to Find Adversarial Instances for Heuristics (25m)

Levente Rácz
Mitigating State Dilution in Axiomatic Proof Trees via Block Attention Residuals (25m)

Jan Jakubuv, Miroslav Olšák, Martin Suda and Josef Urban
Neural Conjecturing for Saturation Theorem Provers (20m)

10:15-10:45 coffee break
10:45-11:50 Chair: TBA Dustin Bryant, Jonathan Julian Huerta Y Munive, Cezary Kaliszyk and Josef Urban
Munkres’ General Topology Autoformalized in Isabelle/HOL (25m)

Miroslav Olšák and David Stanovsky
Project Proposal: Automated Correction of Math Assignements (15m)

Mario Carneiro, Steven D. Miller, Magnus Myreen and Josef Urban
Autoformalizing Fast–Slow Equivalence Proofs for Atlas Computations in the Unitary Dual Problem (25m)

12:30-14:00 lunch and departure from the center

Picture from the conference

Pictures from the previous conferences

Location, Prices and Further Local Information

The conference will take place from August 30 to September 4, 2026, in the CNRS Paul-Langevin Conference Center located in the mountain village of Aussois in Savoy. Dominated by the "Dent Parrachée", one of the highest peaks of La Vanoise, Aussois is located on a sunny plateau at 1500 m altitude, offering a magnificent panorama of the surrounding mountains and a direct access to the downhill ski slopes or cross country slopes in winter. The total price for accommodation and food for the five days will be around 650 EUR.

Arrival/Departure:

The first meal is dinner on August 30st and the last meal is lunch on September 4th. Aussois is less than 2h from the airports of Lyon, Geneve, Chambery, Annecy, Grenoble and Turin. There are trains and buses to Modane from these airports. Aussois is 8km from the Modane TGV station with direct trains from/to Paris. We will organize a bus for the participants from there to Aussois at around 19:10 pm on Sunday, August 30st (waiting for the relevant TGV trains). Further buses to these airports / station can be found here and it is easy to get a taxi from Modane to Aussois and back. We have not yet set the time for the departure of the taxis after lunch on September 4th. This will be optimized based on your departure flights. The center will likely close after our last session on Friday. If you want to stay for the weekend, there are other hotels in and around Aussois. If you have more questions/notes, put them into the registration form and/or look into the FAQ.