Applied problem-solving
A proposed track for building an AI-assisted solution to a defined problem. Evaluation would consider task performance, usability, constraints, and the quality of the participant’s reasoning.
Global AI competitions
A global stage for practical AI capability.
We are developing competition programs that make room for strong ideas, careful evaluation, and the work behind a successful result.
Express competition interest
Build something that stands up to scrutiny.
The goal is to connect learning with a challenge whose rules, evidence, and evaluation can be understood.
Program development: the formats and processes below are proposed. No competition dates, registration openings, awards, or prize commitments have been announced on this website.
Our competition initiative will bring an international perspective to applied AI problems. We want participants to demonstrate more than a compelling presentation or a single favorable result. A useful submission should make its approach understandable, document its limitations, and allow the relevant evidence to be checked.
Model and system evaluation
A proposed track for comparing methods, investigating failures, and improving evaluation practice. Participants would need to explain their data choices and the limits of their conclusions.
Responsible AI design
A proposed track for identifying risks and improving how an AI system is used. The work would connect practical safeguards to a specific setting rather than rely on broad policy statements.
A proposed competition cycle
- Publish the complete challenge.
Define the task, eligibility, permitted resources, submission format, evaluation criteria, and review process before participants begin.
- Support documented development.
Provide clear clarification channels and versioned rule updates. Where a clarification changes expectations, explain how it affects all participants.
- Evaluate and verify.
Assess eligible submissions against the published criteria, review reproducibility evidence, and address questions or conflicts through the stated process.
- Share the learning.
Publish the results and appropriate explanations under the event’s agreed disclosure terms, while respecting participant rights and confidential information.
Integrity belongs in the rules.
| Area | What needs to be clear |
|---|---|
| Scoring | The criteria, weighting, relevant baselines, and treatment of incomplete submissions. |
| Original contribution | Permitted assistance, required attribution, team roles, and use of third-party models or code. |
| Reproducibility | Required code, environment details, data handling records, and evidence of repeatable results. |
| Fair review | Assessor responsibilities, conflicts of interest, clarification opportunities, and result-review procedures. |
Participation with realistic resource expectations
Every future challenge will need to state its computing requirements and access arrangements. Our owned infrastructure may support selected programs, subject to workload suitability, scheduling, and agreed terms. Entrants should not assume that registering interest reserves a GPU or guarantees a computing allocation.
We will consider how differences in prior experience, equipment, and team size affect participation. Track design and eligibility rules should make those constraints visible. Any age requirements, fees, intellectual property terms, and permitted public disclosures will be published before registration opens.
Help shape a meaningful challenge.
Tell us whether you are interested in participating, proposing a problem, contributing evaluation expertise, or supporting a future program.
Discuss AI competitions →