Imagine if we did physics this way. Instead of having scientists study, and carry out experiments, we just got a bunch of punch drunk computer scientists to do thought experiments, and misuse the word fallacy.
Your implication reaches for the stochastic parrot fallacy. You are right about the mechanics, but underestimate the parrot
Take a minute and calculate just how much information is transferred by looking at a word's proximity to near neighbors in a 16,000 dimensional environment.
Here's an thought experiment: How many verbs are near neighbors to the noun "bear"?
What does this tell you about the properties of the noun "bear"?
Same exercise now with nouns and adjectives--what are the near neighbor nouns and adjectives of the noun "bear". Again, what do you learn about the noun "bear" from these near neighbors?
Our brain stores factual information in a very similar way. We don't have a database of "bear" facts in our brain some dude in a little control room looks up when we think "bear". There's no "bear" file in our brains. We have a bunch of thoughts which are near neighbors to the noun "bear" which get activated when we hear/think the word "bear".
From there, it's a recursive cascade.
One near neighbor of "bear" is the verb "lumbering". What are the near neighbors of "lumbering"? What do these near neighbors tell you about a "lumbering bear"?
What does this tell you about other words near to "lumbering" which might also be near to "bear"?
Can two nouns, "bear" and another noun near "lumbering" and "bear" fight each other? Can you use near neighbor words to create a story about these two nouns fighting each other? What are nearest neighbors to "fighting"?
"Godzilla" is a noun which is also near the adjective "lumbering". What would a fight between Godzilla and a bear look like? I'm betting that there is no written work about Godzilla and a bear fighting each other in ChatGPT's training set. Yet, it'll happily construct a story about this for you. That's a new knowledge construct, coming from your "stochastic parrot".
@wastetime@reset_by_peer You're still upset because I called you a shitty reader in response to your failure to properly read what I wrote? Incroyable.
@spilkommen@reset_by_peer Who's coping. I would think that film was awful if generated by a human. It's slop.
It's fairly obvious it's AI (the use of the likenesses of real actors, the inconsistent style, the way it switches film stocks, the inconsistent tone, terrible camera work, etc).
@wastetime@reset_by_peer Okay you're not a 'shitty reader', you just did a really shitty job of reading what I wrote here... Maybe focus on fixing the fault.
@wastetime@reset_by_peer You can't prove any definition about the real world correct. That's literally epistemology 101.
This is why people don't take AI guys seriously - because you don't even know the basics.