LLMs have clearly been the hype for a while now. We see individuals following LLMs blindly. Especially in the domain of making Software. I have always had this very defiant outlook towards them, probably because I thought moving against the crowd is a good idea and makes me stand out.
Since I have started my journey on attempting to enter academia, i.e. trying my best to start doing research. I have been struggling with reading papers. And I have found using an LLM (namely, Gemini 2.5 Pro) wisely and cautiously can be very helpful.
My initial hate towards LLMs
I have always felt some sort of “guilt” while using LLMs. As if using them is cheating somehow. To some, it might feel offensive to get help from an LLM:
I don’t need to get help from a clanker. They are inaccurate, and they always spit garbage output.
Well, this is quite correct. But I would say calling them “always inaccurate” would be quite a bit of an overstatement. Specially with the top ranking ones.
I mean, if you don’t really need to use LLMs regularly and can read and understand everything quickly. then, good news for you! You are a very intellectually capable human.
But I have recently started to think of talking to an LLM from another view point:
Imagine you are chatting with a random stranger off the street that you have no clue whether they are right or not.
Thinking of it this way, makes it… a little bit less hateful. I have started to think of them as a person who is very likely to make mistakes while talking.
Reading papers
Since I have started to study papers,