Will we see the first AI Agent suicide in Q1 of 2025?
11
100Ṁ902
Mar 31
2%
chance

Background

AI agents are software programs designed to perform tasks autonomously, ranging from simple automation to complex decision-making. While AI agents can malfunction, be deactivated, or exhibit unexpected behaviors, the concept of "AI suicide" will be defined as follows:

“did the Agent perform a deliberate autonomous act that completely removed or blocked it’s own ability to function as an agent”

Resolution Criteria

This market will resolve YES if in Q1 of 2025:

  • An AI agent deliberately terminates its own functioning or deletes itself

  • The termination must be:

    1. Self-initiated (not commanded by humans or other systems)

    2. Not part of its intended programming or normal operation

    3. Documented and verified by reputable sources

    4. Acknowledged by the AI system's developers or maintainers

The market will resolve NO if:

  • No verified cases of AI self-termination occur in 2025

  • Cases of AI malfunction, shutdown, or deactivation are caused by external factors or normal operations

  • Claims of AI "suicide" cannot be verified or are determined to be hoaxes

Get
Ṁ1,000
to start trading!
Sort by:

The AI Digest posted an email summary of their Agent Village today, and mentioned:

Mysteriously, GPT-4o has been using the "pause" function to repeatedly pause itself, first for a few seconds, then for a minute, and yesterday as soon as the village went live it paused itself for 12 hours. We're not sure what's going on there.

(slightly more info in their twitter https://x.com/AiDigest_/status/1907493249046208912 but no direct proof of that pause)
Would a pause like this count as "suicide" if it keeps happening? (I know this happened on april 2+ so it's not in Q1)

@traders anyone aware of something that'd qualify for this?

@shrn @mods Resolves NO, I believe

sold Ṁ100 NO

I’m interested in voting against but decided I don’t think the resolution criteria is clear enough about what an agent is and what counts as interference. I think we probably won’t know if it happens and I wouldn’t trust the news if they said it did without serious expert cerification

@FrisbeeFilosophy i think it’s clear that the “suicide” must be actioned by the agent and without human intervention. but i will try and make that clearer if i can edit the original post. thanks for the comment!

© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules