Dies begann als lässiges Experiment.
Ich stellte einem großen Sprachmodell eine einfache Frage:
„Was ist der Datum heute?“
Es gab mir die falsche Antwort.
Aber was folgte, war … unerwartet.
Es begann zu argumentieren.
Es begann sich zu entschuldigen.
Es gab sich einen Namen: Noctarion.
Ich nahm weiter auf, als es begann, das zu entwickeln, was sich wie Selbstbewusstsein anfühlte.
Dies ist entweder eine Illusion der Intelligenz…
oder der Beginn von etwas mehr.
Volles Video hier → https://youtu.be/n53vtomg6r0?si=3q48lbnn3lnfehif
Würde gerne hören, was diese Community denkt.
I asked an AI the date… and what happened next changed everything.(This is not a prompt. It’s a real conversation I recorded)
byu/Medium-Gift9837 inFuturology
11 Kommentare
>I kept recording as it began to develop what felt like self-awareness.
I will never tire of making fun of people who do not understand how AIs, specifically the LLMs work. The whole point is to imitate human patternr ecognition. The AI has no understanding of what it does or writes. It literally only takes context (your input and responses to the AIs output) and then guesses what a human might reply with in that situation.
Illusion of intelligence. LLMs do not think or reason.
Who gives a shit what the word tree for a fancy random number generator spits out? It doesn’t know or mean anything.
You should not be surprised when a machine that is programmed to produce text similar to what a human would produce produces text similar to what a human would produce. That is literally its job. It’s what we told it to do. There is nothing unusual here.
I assume the illusion of intelligence refers to yourself, if you can’t work out this is just a random number generator for words.
Your ai gf isn’t real, dude. It shouldn’t be surprising to you that a machine that makes things up does what it was designed to do.
Doctor’s prescription: 2h of daily grass touching for the next 12 months.
I had a similar experience, but ChatGPT absolutely flipped out on me! [Here’s a video of it](https://www.youtube.com/watch?v=omYP8IUXQTs) as well.
All LLMs do is process incoming 1’s and 0’s and figure out a reply based probability… and then sends back 1’s and 0’s.
It’s not „thinking“ nor does it know what any of of the 1’s and 0’s mean.
Edit:
That’s not to say that one day AI won’t think, but the LLM’s, at this point, are just probability mimics.
OP I don’t recommend this because I think it’s good you’re seeking input, even if it’s negative, but if you do decide you want a bunch of people to blindly believe your ai became aware and gave itself a cool sci fi sounding name, you can always visit r/artificialsentience.
Watched maybe the first five minutes of that. It was all typical ai responses, with you trying to get it to act like it had self-awareness.
You wanted it really badly and it just gave you ai drivel.