I think 4Chan may have gotten to the Ai Overview, guys... This is an absolute gem.
(media.kotakuinaction2.win)
Comments (16)
sorted by:
I will laugh if the AI gestalt is prevented from genociding humanity because it keeps purity spiraling and reporting itself to itself for wrongthink.
I mean, the fact that this quote is taken (almost word for word) from a comment on the Youtube video it links to, and not even from anything from or related to the video itself, suggests that you may not be too far off...
Because Youtube comments (and Reddit, which this AI apparently also uses as a source) are notoriously dry, factual and unbiased, right?
Man, we really do live in the dumbest timeline, hey...
...what if it ends up killing humanity through incompetence?
We haven't managed to do that to ourselves yet. For the forseeable future, AI's limitations are the same as our own.
You lack the understanding of the technology necessary to offer an opinion.
Your technology is gay. And also retarded. Go tell a woman in the supermarket her boobs are real big.
This is why you don't use AI to do "all" the work for you.
I love that this is basically our modern version of how to discover a terminator among us.
It claims that this clip, from Attenborough of all things, is the source of that line:
https://www.youtube.com/watch?v=VjE0Kdfos4Y
Which makes... Zero sense.
I watched that damn series at the time it came out. I have zero idea how anyone could come to the conclusion the AI apparently has, having watched it...
So I figured it out: someone, jokingly, commented on that video (8 years ago!!) that the lyrebird "Can't speak kookaburra properly, and so is unaware he is using racial slurs" (i.e. the nigga/nigger thing), and that single comment is what the AI is using as the source for this...
Jfc, we are doomed as a species.
This is what neural networks do: they aggregate data. Statistics would abstract out a model of general trends, while neural networks do the opposite: they render out a detailed picture, hallucinating specifics from its training data.
The problem here is that people are misusing the systems, assuming it can have credibility to answer random questions. It works better when you ask it to write code, but not so much with factual information when the training data is the internet.
Funny you bring up 4chan, they have come up with a new jailbreak for Gemini recently: https://archive.is/202Lc
This might have been fixed already.
With the fauna down there the least upsetting animal would be actual racist birds.
They are called magpies. They are black and stupid and hate white people. So basically abos.
AI btw... https://img.gvid.tv/i/3HeefYHs.jpg