Skip to the findings

FRONTIER THOUGHT / FINDINGS AND NEXT STEPS

THE AI CHOICES

Better AI does not automatically mean better work or stronger learning. The useful test is what a person can do better, with responsibility and correction still clear.

The discussion so far22 November 2026

THE QUESTION WE STARTED WITH

Can society stay in control of systems it increasingly depends on?

WHAT THE SOURCES TELL US

Useful intelligence and autonomous control are separate design choices. An essential service also needs people who can detect failure, intervene and continue the work.

THE NEXT QUESTION

If an AI system fails in a Greek hospital or municipality tomorrow, who can stop it and keep the essential service running?

For a hospital or municipality, resilience means knowing which essential task must continue, who may intervene and how long restoration takes. Bletsas’s interview makes this a concrete Greek operational question.

Organizing a discussion on this question?

Use these findings and sources as the starting point for a panel, interview, email exchange or small roundtable.

Coming next: A real AI gain: who gets the time back? ↓

HOW THE ANSWER DEVELOPS
  1. 01The starting point

    Useful intelligence and autonomous control are separate design choices. An essential service also needs people who can detect failure, intervene and continue the work.

  2. 02Further evidence

    NIST guidance recommends rehearsed fallback and clear deactivation responsibility. CaMeL research demonstrates a boundary between trusted instructions and untrusted content within a specified threat model. Test blocked actions separately from misleading advice and continuity when the model stops.

    Still open Voluntary guidance and a benchmark do not certify a service. Permission design, user approval fatigue and actual recovery still need local testing.

  3. 03The decision to test

    Before the service goes live, agree who can pause it, how people get help and what must be checked before it restarts. Test what happens if the decision maker or the necessary logs are unavailable, using synthetic records.

    Still open Contractual authority, applicable duties, log access and staff availability need confirmation. The proposed exercise has yet to be run, and the operator’s legal authority remains to be checked.

WHO CAN HELP ANSWER THIS?

The people closest to the problem.

Hospital operator or municipal service manager

Which essential task must continue during an AI or cloud outage, and who owns that decision?

Continuity procedure, named duty role, manual fallback and last exercise record.

Cybersecurity lead and system supplier

What signals trigger an investigation or shutdown, and who can see the record?

Incident categories, audit access, containment procedure and measured detection and restoration times.

Frontline worker and affected service user

How do you challenge a harmful decision when the system or supplier is unavailable?

An actual escalation route, a response deadline and a completed resolution example with personal details removed.

Each perspective can fill a different gap in what we know.

Sources and what remains uncertain

Read the original work alongside the analysis. For the local case, we still need the evidence requested above.

  1. Yoshua BengioThe catastrophic risks of AI — and a safer path ↗8 April 2025 · bengio-ted-20250408
  2. Michalis BletsasInterview on cyberattacks and national resilience ↗4 November 2026 · bletsas-resilience-2026
  3. NISTArtificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile ↗July 2024 · nist-genai-profile-202407
  4. Edoardo Debenedetti, Ilia Shumailov, Tianqi Fan, Jamie Hayes, Nicholas Carlini, Daniel Fabian, Christoph Kern, Chongyang Shi, Andreas Terzis, Florian TramèrDefeating Prompt Injections by Design ↗24 June 2025 · camel-prompt-injection-20250624

NEXT DISCUSSION

23 November 2026 · 09:00 Athens time

A real AI gain: who gets the time back?

Before expanding one Greek AI service, what evidence from complete cases would show a benefit after checking and correction, and who should receive the saved time?

WHERE WE BEGIN

The model may improve faster than the organisation. Compare Amodei on adoption, Nadella on the operating environment and Mitsotakis on visible benefits to people, using the workplace studies as the evidence test.

THE PRACTICAL OUTCOME

One page comparing the service: complete case time, correction work, user completion, omitted cases, benefit recipient and a clear rule for deciding whether to stop or expand.

The published perspectives behind this question

A WORKING BRIEF, READY TO EDIT

Make this the next discussion.

The finding, original sources and evidence requests come with you. Review the question and choose the people who can clarify it.

Saved briefs are visible to authorized site editors. Use public material or material approved for that workspace.

Organizer sign in to save ↗