The MAIA BRAIN scientific challenge
Quantum Leap. Put a proof on a qubit.
Neurosymbolic AI reasons by rules and proves what it concludes. Quantum hardware searches spaces no classical machine can walk. We are offering CHF 10,000 for the work that joins the two without losing the proof.
-
CHF 10,000
One prize, for the submission that moves provable reasoning furthest onto quantum hardware
-
Four tracks
Search, constraints, learning and verification. Enter one, or all four
-
Open entry
Universities, laboratories, companies and independent researchers
-
Judged on proof
A result counts when it can be checked, not when it can be shown
The question
Can a machine that works in probabilities return an answer you can prove?
Every quantum computation ends in a measurement, and every measurement is a sample. MAIA BRAIN is built on the opposite promise: the same question gets the same answer twice, with the rule that produced it.
The hardware may guess. The answer may not.
The interesting ground lies between the two. Symbolic reasoning spends most of its effort searching: for a proof, for an assignment that satisfies every rule, for the one plan that breaks none of them. Search is where quantum hardware is strongest. Checking a proof, once found, is cheap and classical.
So the challenge is this. Let the quantum machine find. Let formal reasoning check. Show us a result that is faster, larger or cheaper than the classical route, and that a sceptic can still verify line by line.
The four tracks
Where a quantum machine could earn its place in reasoning.
Enter in one track or across several. The prize goes to the strongest submission overall, not to the best in each.
-
01 Search over proofs
Proof search and rule chaining expressed as a problem quantum hardware can sample, with every candidate the machine returns checked by a classical verifier before it counts.
-
02 Constraints at institutional scale
Satisfiability, scheduling and allocation under thousands of business rules at once. Show where an annealer or a gate-based machine finds the assignment that breaks none of them.
-
03 Learning rules, not weights
Inducing readable rules from examples, with the quantum step inside the search for the rule. What comes out has to be a rule a person can read and sign.
-
04 Verification and certainty
Certificates for answers a quantum machine produced. How do you prove that a sampled result is the right one, and how much classical work does that proof cost?
How it is judged
The bar is a proof, not a demonstration.
A result on a simulator is welcome. A claim of quantum advantage without a classical baseline is not.
-
Checkable
Every answer the submission produces can be verified by formal reasoning, independently of the hardware that found it.
-
Reproducible
Code, data and the measurement protocol are shared, so a second team reaches the same conclusion from the same inputs.
-
Honest about the hardware
Noise, depth and qubit counts are reported as measured. A projection is labelled as one, and compared against the best classical method.
How it runs
Four steps, from interest to award.
The full rules, the timetable and the judging panel are published on this page before submissions open. Register your interest and we will send them to you first.
-
01
Register your interest
Tell us who you are and which track you have in mind. One conversation, no forms to file.
-
02
Receive the brief
The full rules, the benchmark problems and the verification harness every submission is checked against.
-
03
Submit the work
A paper, the code and the evidence. Hardware runs, simulator runs, or both.
-
04
Review and award
Each submission is re-run and its answers checked. The prize of CHF 10,000 goes to the work that moves provable reasoning furthest.
Questions
What people ask first.
Do I need access to a quantum computer?
No. Simulated results are accepted, provided the simulation is stated as one and its limits are reported. Runs on real hardware weigh more when the rest is equal.
Who owns the work?
You do. Entering the challenge does not transfer your intellectual property. The terms of publication are set out in the full rules.
Why is a company running a scientific challenge?
Because the question is open, and an open question is answered faster in public. MAIA BRAIN reasons by rules it can prove. We want to know how far that can travel onto the next kind of hardware.
When does it open?
The timetable is published here with the full rules. Register your interest and you will hear the dates before they are announced.
The next step
Bring the hard part.
If you work on quantum algorithms, formal methods or both, tell us what you would attempt. We will send you the brief as soon as it is published.