[jryb]

LLMs may be enabling, but OP explicitly stated "I wanted to dig deeper into the subject, but not by reading a boring textbook", which, of course, would have eliminated the issues in this tool, or at least made it clear to them that they needed to dive deeper before publishing. I feel like there might be some analogy with blaming cars for drunk driving - by definition, not possible without cars, but you can drive responsibly if you choose to.

[jart]

That's how Plato taught Aristotle. I'd much rather have a dialog than read a textbook. You just have to think critically and fact check. You can't just trust whatever the robot says because it interpolates knowledge.

[jryb]

Sure, but there apparently wasn't enough of a dialog either. With a textbook, you're confronted with facts and explanations that you didn't ask about, or even knew to ask about. Don't get me wrong - my original recommendation to take an interactive course is still the best option in my mind, as simplifications made for the benefit of the learner often lead to apparent contradictions that an instructor can clarify. But at some point you do just need the set of raw facts to be able to work with these systems.

[jart]

Only genuine socialization does that, since you can flip through a textbook. It's the main reason why people go to universities. To hear all the answers they didn't think to ask. LLMs actually do that a little bit. It's both a wonderful and terrifying prospect, since there's so much risk with that kind of feature to introduce bias, but it's great when it works.