you are viewing a single comment's thread
view the rest of the comments
[–] 55 points 10 months ago (5 children)

I am guessing the reporter wanted to remind people tools exist for this, however the reporter isn't tech savvy enough to realize ChatGPT isn't one of them.

  • source
  • parent
  • hideshow 5 child comments
  • [–] 27 points 10 months ago* (4 children)

    afaik, there actually aren't any reliable tools for this.

    the highest accuracy rate I've seen reported for "AI detectors" is somewhere around 60%; barely better than a random guess...

    edit: i think that way for text/LLM, to be fair.

    kinda doubt images are much better though...happy to hear otherwise, if there are better ones!

  • source
  • parent
  • hideshow 4 child comments
  • [–] 17 points 10 months ago (3 children)

    The problem is any AI detector can be used to train AI to fool it, if it's publicly available

  • source
  • parent
  • hideshow 3 child comments
  • [–] 12 points 10 months ago* (2 children)

    exactly!

    using a "detector" is how (not all, but a lot of) AIs (LLMs, GenAI) are trained:

    have one AI that's a "student", and one that's a "teacher" and pit them against one another until the student fools the teacher nearly 100% of the time. this is what's usually called "training" an AI.

    one can do very funny things with this tech!

    for anyone that wants to see this process in action, here's a great example:

    Benn Jorda: Breaking The Creepy AI in Police Cameras

  • source
  • parent
  • hideshow 2 child comments