For the past couple of months I have seen more and more screenshots of people asking intentionally stupid, culturally relevant questions to Google AI. The answers are actually entertaining with the AI "being in" on the joke.

I am certain these answers have been encoded, asked, screenshot and posted to Twitter/Reddit, by AI companies themselves in order to create more goodwill towards AI and remove memories of the hilariously bad early AI Google answers.

all 5 comments

sorted by: hot top controversial new old
[โ€“] 4 points 2 days ago

I can't say I've ever seen one of these things that you're talking about.

  • source
  • [โ€“] 1 point 1 day ago (1 child)

    Might want to give an example.

    You mean like "Open the pod bay doors, Hal" "I can't do that, Dave*? The AI does that because it's been fed that exchange a gazillion times so the most likely "correct" response to "open the pod bay doors" is "I can't do that". Same with other culturally significant texts.

  • source
  • hideshow 1 child comment
  • Probably also sillier things, too. Stuff like "I'm a fish and I see a hook with a worm on it" and the ai will be like "Oh no! Swim away now! Do not try to bite the worm! Its a trap and you will definitely be going on an adventure that you hadn't bargained for if you decide to bite that worm."

  • source
  • parent
  • this may be analog to how you can goog search recursion. and it spits out recursion.

    if it was then yeah, i would say it is a [learning] plant for goodwill.

    otherwise, ai can easily pick up memes so it could have been just the natural flow of info

  • source
  • [โ€“] 1 point 2 days ago

    Haven't seen these either. I do agree, if they are primarily in spaces with lots of promotional content then it's more obvious.

  • source