The trick that AI pulls is just that it is really good at appearing competent
Yeah it’s almost like the AI is an unthinking machine trying to model what a person would answer to the question (prompt) provided to it.
I’m so goddamn tired of people (not you) acting like an LLM “knows” anything nevermind whether it’s right or wrong or could even tell you if it didn’t know the answer. It’s not like a person that can say, “I don’t have that knowledge so I know any answer is likely wrong” the machine is going to give you an answer every time. That answer will never be “idk” it’ll be whatever hallucination it’s confident enough is right based on the model and training.
I do think there are use cases for it and the technology is cool but the vision being sold and propping up our economy with a venture capitalist circle jerk is so far removed from reality.
I’m pretty sure the fucking economist you’re replying to is not “teach(ing) something different from the mainstream” that’s the whole point of their post.
The whole point is the AI tools are hallucinating* completely incorrect answers and that students trying to use them as study aids and tools to learn with are going to have a lot of fake and wrong information to sort out and good luck to them knowing what is or isn’t incorrect or wrong when they’re just trying to learn a subject they know nothing about.
The only reason you or I or anyone else knows when AI hallucinates in its output is because we know stuff from our prior knowledge and can catch it that way. If you’re ignorant to a subject and trying to learn stuff you’re not going to know off hand the AI is lying to you if it sounds convincing enough for a layperson to accept.
*which I kind of hate that term because to the AI there’s no difference between truth and hallucination - it’s just output as far as the robot is concerned