The event
- Name
- Quantum AI Conference 2026 (QNLPAI 2026)
- Dates
- 14 to 16 August 2026
- Venue
- IU Indianapolis Campus Center, 420 University Boulevard, Indianapolis, Indiana (registration and opening reception at Hine Hall, banquet at the Madam Walker Theatre)
- Host
- Indiana Indianapolis
- Also on the programme
- Damir Cavar (Indiana, conference chair), Bob Coecke (Relational Intelligence, section chair for Quantum NLP), Keeper L. Sharkey (ODE), Jasmine Brewer (Quantinuum), Rickey Carter (Mayo Clinic in Florida)
- Programme
- qnlp.ai
Trip 14 to 16 August 2026 · Written up 16 August 2026 · Last expanded 29 September 2026
Brutus and I were in Indianapolis, where I presented my research at the QNLPAI poster session, hosted at IU Indianapolis and the Luddy School.
The project asks a deceptively simple question: when a Transformer succeeds on a logical task, how much of the apparent reasoning was learned by the model, and how much was supplied by its representation (inspired by Bob Coecke’s pioneering paradigm), its action interface, its search procedure, and its verifier?
I got thoughtful feedback and will keep developing the paper this fall alongside larger and more demanding experiments. Compute permitting, we may borrow some additional cluster power from GT.
The branch that did not die
I arrived on too little sleep and was not sure what to expect, and it turned out to be decidedly foundational.
The common theme across the conference was a rigorous, compositional programme for QNLP that reconnects to the generative roots of natural language processing, mapping syntax to vector and diagrammatic semantics while staying aligned with neuro-level constraints.
For anyone outside the field, that sentence carries more freight than it looks. NLP had two branches: one was championed by Noam Chomsky, and although a great deal of work was done there it has mostly withered relative to the second, the distributional and statistical approach that bore enormous fruit through the big-data decade and culminated in the LLM. Seeing the first branch make a credible return, by this route, was a surprise.
Anyone who has studied both quantum mechanics and machine learning will have noticed the similarity in the mathematical apparatus. I never pursued the link seriously, and several researchers here are doing exactly that, and doing it properly.
Who I met
- Bob Coecke. A very lively dinner on process and diagrammatic languages, and how they sit against quantum and logical abstractions. Bob helped join quantum mechanics and computing from a theoretical purview using category-theoretic languages. We also got into how the foundations of mathematics and science have shifted over the past twenty years, which is not a conversation you get to have often.
- Enan Srivastava. A fellow Buckeye working the same quantum and AI intersection. We had to do the O–H. There are at least four of us now, which means we can finally spell the whole thing.
- Keeper Sharkey, Taha Selim and Sparsh Srivastava, all at the same dinner. I got to share some of my own research from the past few months, which was useful mostly for finding out which parts of it I could not yet explain quickly.
Where the idea came from
The provenance is worth recording, because it is not glamorous: I got steered toward this line while talking to someone in the IU Bloomington Student Union cafeteria last year, while they finished a plate of cheese fries. Inspiration comes in strange places.
A question for the Ohio State people reading this: what is our next move in quantum information and AI?
Photographs
Links
Poster, slides and the QNLPAI programme. To be added.
Open entry
This page is not finished, and is not meant to be. I add photographs, people and links as I dig them out, so it keeps growing after publication. Why the whole site works this way is on the colophon.
