Yeah it's a weird mix of issues with the backend model and issues with the CLI client and its prompts. What makes it hard for them is the teams aren't talking to each other. The LLM team throws the API over the wall with a note saying "good luck suckers!".
You'd hope AI would be used more to support children and teach them. Can you imagine a patient teacher who's available 24/7? I actually ask LLMs to teach me stuff sometimes, and it does work, but... early days.
Possibly a distinction needs to be made between raw llm output, raw google output (like lmgtfy), or any other tool's raw output on the one hand, and a synthesis of your conclusions after having used these tools together, on the other.
Obviously cut&pasting the raw output of a google search or pubmed search or etc would be silly. Same goes for AI generated summaries and such. But references you find this way can certainly be useful.
And using spelling checkers, grammar checkers, style checkers, translation tools or etc (old fashioned or new AI-enhanced) should be ok if used wisely.
Taking ownership isn't the worst instinct, to be fair. But that's a slightly different formulation.
"People are responsible for the comments that they post no matter how they wrote them. If you use tools (AI or otherwise) to help you make a comment, that responsibility does not go away"
You know how some people can hardly find the back of their own hands if they googled them?
And then there's people (like eg. experienced wikipedians doing research) who have google-fu and can find accurate information about the weirdest things in the amount of time it takes you to tie your shoes and get your hat on.
Now watch how someone like THAT uses chatgpt (or some better LLM) . It's very different from just prompting with a question. Often it involves delegating search tasks to the LLM (and opening 5 google tabs alongside besides) . And they get really interesting results!
I actually use LLMs to help me dig up the sources. It's quicker than google and you get them nicely formatted besides.
But: Just because it's easy doesn't mean you're allowed to be lazy. You need to check all the sources, not just the ones that happen to agree with your view. Sometimes the ones that disagree are more interesting! And at least you can have a bit of drama yelling at your screen at how dumb they obviously are. Formulating why they are dumb, now there's the challenge - and the intellectual honesty.
But yeah, using LLMs to help with actually doing the research? Totally a thing.
You could instead ask Kimi K2 to demolish your point instead, and you may have to hold it back from insulting your mom in the ps.
Generally if your point holds up under polishing under Kimi pressure, by all means post it on HN, I'd say.
Other LLMs do tend to be more gentle with you, but if you ask them to be critical or to steelman the opposing view, they can be powerful tools for actually understanding where someone else is coming from.
Try this: Ask an LLM to read the view of the person you're answering to, and ask it steelman their arguments. Now think to see if your point is still defensible, or what kinds of sources or data you'd need to bolster it.
Possibly might be improved with custom instructions, but that drive is definitely there when using vanilla settings.
reply