Ongoing
UK AI Salary & Skills Survey
What AI practitioners earn, and what employers pay for.
Progress62%
IoAI Labs · Independent research · Open methodology
Research in the open,
on the Institute's terms.
IoAI Labs is the Institute's independent research arm. We study how AI is practised across the UK, design the tools and frameworks our members need, and partner with frontier labs on a shared goal: making AI more accessible and better for everyone.
— How the Labs work —
Designed with members,
not for them.
Everything the Labs do is built in partnership: with the practitioners who use AI every day, and with the frontier labs building it.
01
Built with the membership
Our members set the problems, shape the solutions, and test every tool and framework we design in real work.
02
Alongside the frontier labs
We partner with the labs building the models: early access, joint experiments, and shared work on using new capabilities well.
03
From research to practice
Findings don't sit in papers. They become the tools, frameworks and guidance practitioners can actually pick up and use.
— Active research —
What the Labs are
working on now.
Research and solution designs in progress, published as they develop rather than only when they are finished.
Ongoing
Orchestrating a Fleet of 1,000+ Agents
A reference design for coordinating agents at scale.
Progress55%
Active
State of AI Practice
How UK organisations are really putting AI to work.
Progress45%
Open
Frontier Model Safety Collaborations
Working with the labs on responsible evaluation.
Progress40%
Recruiting
Open Evaluation Harness
Open-source tooling for members to test models on their own work.
Progress33%
Recruiting
The AI Career-Switch Report
The routes people take into AI, and what actually lands them the job.
Progress28%
Open
AI in the Public Sector
What responsible AI deployment looks like in public services.
Progress20%
— Working with frontier labs —
Alongside the labs,
not against them.
We partner with frontier labs because we want the same thing they do: AI that reaches more people and works better for them. That shared goal, rather than a scoreboard, is where the relationship starts. It also opens up early, privileged access to new models and safety tooling, including advanced cyber-safety evaluation, which we put to work on real problems alongside the labs themselves. And the rankings on our leaderboards come from our members, not our staff: accredited practitioners judging models from their own work.
- Privileged access to new models and safety tooling, including advanced cyber-safety evaluation
- Joint experiments with the labs, run on real practitioner workloads
- Member-run reviews, not staff-scored, feeding our public leaderboards
— Publications —
What we publish,
free to read.
Reports, surveys, position papers, consultation responses and methodology notes from the Labs.
Position Paper
12 pp
When AI Becomes Too Capable to Release
2026
Position Paper
7 pp
Vera Rubin and the Future of AI Infrastructure
2026
Position Paper
12 pp
Rethinking Responsibility
2026
Position Paper
8 pp
AI's Infrastructure Boom
2026
Position Paper
9 pp
Germany's AI-Driven Defence Overhaul
2025
Position Paper
8 pp
Ireland's AI Watchdog
2025
We do not run a lab to publish papers. We run a lab to make AI work better for the people practising it, and we publish everything so anyone can check our working.
Put your next question
to the Labs.
A short brief to the Labs team. Collaborations, model reviews and publication requests answered within two working days.
— Go deeper
— Insights
The editorial side
Commentary, news and published research from the Institute, free to read.
Read Insights— Consultancy
Need it applied?
Independent, standards-led advisory for organisations putting AI to work.
See AI Consultancy— Partnerships
The full picture
Every way the Institute works with organisations, universities and labs.
All Partnerships