19 votes

We need to control AI agents now

2 comments

  1. MimicSquid
    (edited )
    Link
    The solutions feels a little half-baked. A lot of the section talking about the concerns discuss malice, but the solutions section only proposes solutions which are opt-in. Much like "do not...

    The solutions feels a little half-baked. A lot of the section talking about the concerns discuss malice, but the solutions section only proposes solutions which are opt-in. Much like "do not track" options for website cookies, developing systems that only work when everyone agrees to be nice ignores the fundamental challenge of a world wide web where there will always be people who choose to do things that aren't.

    EDIT, to open up the conversation instead of just being negative: Does anyone actually have a solution for this that accepts the existence of bad actors? It seems like a lot of the things that would rule out bots are things that have been resisted in the name of online anonymity: required ID/real name connection to your accounts, connecting it to other real life information, etc. Proof of personhood, in one way or another.

    6 votes