IIDA Focused Research Organization
Get involved

Research

Four tracks. One survives.

Year One runs four working groups in parallel. Each has written kill criteria, a named artefact, and a lead position that is currently open. At the fourth-quarter gate we commit to one focused mission.

Year one

Track 01

8–12 contributors

Seeking lead

AI Reasoning with Game Theory

Reasoning is the next frontier in AI, and game theory offers rigorous foundations for multi-agent decisions, cooperation and adversarial dynamics. Most reasoning benchmarks still measure a single model answering a static question. Almost none measure what happens when two systems want different things.

Kill criteria. If the benchmark does not separate models that a practitioner would already rank differently, the track closes at the Q4 review rather than being carried into Year 2.

Read the track brief →

Track 02

6–10 contributors

Seeking lead

World Models and JEPA

Joint Embedding Predictive Architectures point past purely generative AI toward sample-efficient learning, grounded reasoning and world models for embodied agents. For a research programme with limited compute, an architecture family whose central claim is sample efficiency is not a fashionable choice — it is the practical one.

Kill criteria. If replication does not reach published baselines within two quarters, the track narrows to the Indic low-resource question only, or closes.

Read the track brief →

Track 03

6–10 contributors

Seeking lead

UX and UI in Agentic Applications

AI is moving from chat to ambient, agentic and multi-device. The interaction layer — how a person delegates, oversees, interrupts and recovers — is where adoption and trust are actually won or lost. It is also the layer with the least published, reusable research.

Kill criteria. If the pattern library is not being cited or forked outside IIDA within two quarters of release, it becomes a maintenance-only artefact rather than an active track.

Read the track brief →

Track 04

6–10 contributors

Seeking lead

Tokenomics — Valuing AI Token Usage

As language models become infrastructure, the unit economics of inference remain poorly understood. Pricing, fair use and per-token value attribution are open problems with genuine policy weight — and they are a live blocker on enterprise adoption, which makes them squarely an SME problem.

Kill criteria. If the attribution toolkit cannot reproduce a finance team's own cost model on a real workload, the framework is wrong and the track closes rather than being defended.

Read the track brief →

The plan

The narrowing is scheduled, not hoped for.

The Year One convergence gate Four research tracks run in parallel through Year One, narrow into a single committed mission at the fourth-quarter gate, and that single line thickens through Years Two and Three as output scales from three papers to twenty. Reasoning & game theory World models · JEPA Agentic UX Tokenomics Q4 gate commit to one mission Year 1 · explore · 3 papers Year 2 · accelerate · 10 Year 3 · scale · 20
Solid — funded and running Dashed — subject to kill criteria Weight — output volume

Wider topic pool

Not yet tracks

Live in the pool, waiting on someone.

These came out of member proposals and have not been started. An empty slot with a name on the door recruits better than a full-looking list.

TopicStatusIntended artefact
Compliance for SMEs Open — seeking a lead Possible self-serve product
Datasets for EV, solar and renewable energy Open — seeking a lead Open dataset
Protocols for human–agent collaboration Scoping To be defined
Fraud detection across unstructured documents Scoping Whitepaper and trial MVP

How work moves

Workflow

Including the exit that isn’t a paper.

  1. Proposal sourcing from researchers, industry and members
  2. Initial screening for relevance, novelty and feasibility
  3. Shortlisted ideas become research topics under interim governance
  4. Problem definition, literature review, methodology, initial analysis
  5. Periodic interim reviews on progress and originality
  6. Low originality exits here — redirected to a blog, point of view or whitepaper rather than carried as research
  7. IP disclosure, patentability and prior-art check, filing decision
  8. Commercialisation decision — licence, spin-off, government adoption or sale
  9. Execution, revenue sharing and portfolio maintenance

Step six is the one that matters. Most research programmes have no honest way to stop, so weak work gets carried until it embarrasses someone. Full governance and IP framework →