Publish the follow-up
Each commitment with an owner, a date and a public status. Test: for one AI commitment made by your government, can you find who owns it and what its status is?
NO. 3 · 2026-10-07 · FOR POLICY AND TECHNOLOGY
QUESTION OF THE WEEK
One open question at a time, with a source to check and room to disagree.
A global dialogue on AI governance now exists. How will anyone know whether what is said there changes anything?
WHY IT MATTERS
Talking together is a first step; change shows up in follow-through. Who would check, with what evidence and in what form is an open question, and it is ours, not the source’s.
WHAT THE SOURCE SAYS
- The UN Global Dialogue on AI Governance is described as the first platform under a UN General Assembly mandate where Member States and all relevant stakeholders meet on AI governance.
- Its inaugural session took place on 6–7 July 2026 in Geneva.
- The joint secretariat is made up of the ITU, UNESCO, the UN Office for Digital and Emerging Technologies (ODET) and the Executive Office of the Secretary-General; the ITU and UNESCO coordinated the first session.
UNESCO · UN Global Dialogue on AI Governance opens in Geneva ↗
THREE POSITIONS TO TEST
These are our working hypotheses, not claims by the source.
Include those affected
Residents, workers and small organisations at the table, not only institutions. Test: who from your community could take part, and how would they find out?
Small, checkable steps
A few measurable things a year rather than many broad ones. Test: which statement from a global meeting could you check in your own city within six months?
YOUR ANSWER
Answer in the Courtyard in a few lines, or describe how you would test one position in a small pilot. No registration, no personal contacts.
EARLIER QUESTIONS
NO. 2 · 2026-10-07 · FOR POLICY AND FORESIGHT Why do governments write digital strategies better than they carry them out, and who should check?
Why do governments write digital strategies better than they carry them out, and who should check?
WHY IT MATTERS
A plan on paper is not a service that works for a resident. The gap between the two is where trust is lost, and it is rarely measured in public.
WHAT THE SOURCE SAYS
- The OECD Digital Government Outlook 2026 covers 36 OECD members and 8 accession candidates; the average Digital Government Index score rose by 9 percentage points and the open-data index (OURdata) by 5.
- It finds that governments are much better at developing digital strategies than at implementing or monitoring them.
- Only two-thirds of countries report widespread adoption of data interoperability systems, and digital identity still fails to guarantee seamless access to public services.
OECD · Digital Government Outlook 2026 (summary on the European Digital Skills and Jobs Platform) ↗
THREE POSITIONS TO TEST
These are our working hypotheses, not claims by the source.
Measure the doing
Publish what was delivered, not only what was planned. Test: for one plan in your city, can you find a public report of what was actually delivered?
Let residents monitor
Simple public dashboards and a way to report that a service does not work. Test: who can report a failure, and does anyone answer?
Promise less, test more
Short pilots with a fixed end and honest notes about what failed. Test: which plan could be cut to a six-week experiment?
NO. 1 · 2026-10-07 · FOR PHILOSOPHERS AND SCIENTISTS Can AI widen who takes part in decisions without replacing real participation?
Can AI widen who takes part in decisions without replacing real participation?
WHY IT MATTERS
Governments try AI for translation, transcription, moderation and simulation, but struggle to scale it. The same tool can let thousands of neighbours be heard, or replace the meeting by a model of the meeting.
WHAT THE SOURCE SAYS
- The OECD review (June 2026) describes nine kinds of AI use in citizen participation, from translation and transcription to moderation and simulation.
- It warns about skewed data, opaque decisions and simulations that replace real engagement.
- It finds that governments pilot a lot but struggle to scale, for lack of skills, coordination and practical guidance.
OECD · Artificial Intelligence and the Future of Citizen Participation ↗
THREE POSITIONS TO TEST
These are our working hypotheses, not claims by the source.
Widen
Translation and transcription let more neighbours join in their own language. Test: how many more people took part, and who is still missing?
Be careful
A simulation of residents is not residents. Test: would you accept a decision “made” by a model of your own house?
Set conditions
Use AI only where there is clear evidence that it meets user needs and improves outcomes, and publish that evidence. Test: what evidence would convince you?