"AI" programs like chatgpt have zero intelligence going on, they do not know anything, they have never had a thought, they don't understand the things they say, they will tell you false information, they are not capable of verifying anything they tell you
Comments
In a sense, its worse than what you said.
It only "knows" what you prompt it with and the rest it guesses.
Yea, it doesn't reason. At all.
They are all statistical models of established processes and events, proving the most likely answer to a specific change.
But calling it AI was the worst marketing gimmick ever.
Since talking is the only genuine skill LLMs have, they make great second language convo partners. And a small model like that could be made with a public domain corpus and low carbon footprint.
We're data scientists. We train AI models. They'd replace us with a magic 8ball if it was as confident as chatGPT.
People attempting to justify their outrage would post chat GPT summaries of her thesis, then get all outraged and make threats and wail and so on based on what chat GPT said she said.
So the question of what ChatGPT was presenting is sort of interesting, right?
Sort of just... they asked it to write something that would enrage them and, based on essentially nothing, it complied.
As for giving false information, they do it much fewer times than they give right (not true) information, that’s how they pass bar exams; ergo they’re capable of verifying
that information is pervasive and,
it is easily expressed in text and,
you can verify the information.
Statistically speaking, humans are morons
So you are getting statistical bullshited
These motherfuckers killed the passion I had for tech and I will *never* forgive them for it.
*Well*, some recent models like GPT4o are able to cite their sources, and you can go check those.
Never blindly trust an LLM output, but an LLM that can cite its sources is a powerful tool (like Gemini Deep Research).
The output is so inconsistent that you'd have to verify its claims against wikipedia (or better, an actual source!), which you could have done anyways and skipped the AI.
What did the AI do except unknowingly give you bad information?
Take a Big Mac
Mash them together
"Imma chef now"
AI lies without any reason, and if you don't check it from a real source (invalidating the AI) there's no way to know when.