[cmrdporcupine]

> If, on the other hand, LLMs help you with making much faster progress in understanding the subject that you study

My experience is that they uncomfortably do both. You can "understand" something conceptually quicker -- like you have a new brain-muscle-thing that lets you cut through the hard difficult tedious corners to get to the meat of the matter.

But then you also can become reliant on it, and have difficulty doing the mechanistic rote work of working through it yourself.

Like the really big powerful calculator that it is, really.

[SecretDreams]

It's two fold. They're learning and understanding more things, but at a very surface level and without the nuance and ability to actually use the knowledge because they have none of the muscle memory and hard work associated with learning it.

You can use AI or the internet to learn the basics of how a gas engine works in a couple of minutes. But you'd be incapable of actually working on a gas engine or designing one.

Surface level knowledge gets you surface level functionality. You don't become good at something from surface level knowledge, but you might think you're good at it.

[hirvi74]

Perhaps it was merely my degree or program, but I do not feel like I learned a lot more than surface level knowledge.

I have an accredited degree in computer science. But what does that truly even signify? For example, even with my 10 years of experience coupled with my degree, I could not create a programming language from scratch without some tutorial, nor could I create an OS, a compiler/interpreter, a basic CPU, RAM, any secondary storage, etc.. Sure, I could tell you about basic digital logic, basic OS principals, basic algorithms in a poorer explanation than a Wikipedia article, etc..

To use your gas engine analogy, if you dropped me on deserted island with all the parts I needed to make a basic, simple computer I would unlikely be able to do so with out some sort of in-depth guide. I am not talking about slotting mass produced parts on a motherboard. Rather, something like this:

https://www.instructables.com/Build-a-Computer-W-Basic-Under...

[cmrdporcupine]

If used correctly though I think the models in fact can be useful for gaining depth. With the right prompting they can actually perform a pedagogic role. One must just resist the temptation to have them do the work for you.

I've used my phone taking pictures + Codex + a PDF of my tractor manual to help me effectively diagnose and manage repairs in my tractor. (Though these models remain terrible at the physical world, getting physical orientations wrong, front back etc. Much like myself)

Likewise I had Gemini help me tear down my mower's carburetor and diagnose issues there.

(So much so that I've wondered about building some kind of "shop buddy" -- some kind of durable laptop and set of cameras ... on a cart. Running models that have access to manuals and cameras and TTS and voice input? "Hey, shop buddy, look at this fuse and tell me what is before and after it in the electrical system.")

This is helping me learn and do something I couldn't really effectively do before by walking me through steps.

My youngest has had Gemini write math questions for them, to help study. Not do the math, but write questions.

In the end it comes down to prompting, like everything.

Which makes me wonder if the answer for higher education is just to provide the students with specific coding agents they're specifically allowed to use -- ones that would push the student through problem solving and working on the problem together.

[SecretDreams]

> One must just resist the temptation to have them do the work for you.

We are in the instant gratification era of humanity where a dopamine rush drives most people. This is a systematic shift that happened through the introduction of smart phones and social media and then progressed for a good decade to what we have in front of us today.

Asking people to "resist the urge" when they've been programmed/brought up to feed the urge is not pragmatic unless you are also proposing a way to erase the damage done from the instant gratification era.

We're in the end game presently. For every one person like you and your examples, there's gotta be 100x or more who are not using the tools the way you've presented them.

[cmrdporcupine]

I'm no saint. For coding projects I am absolutely in the same boat as everyone else.

My other examples have to do with current limitations of the tools. Obviously there's no Claude Code for Meatspace that just takes over and does things for you. (Yet)

What I'm trying to point out is that the tooling has been made this way on purpose and I agree substantially with your point. But I also think human agency is involved. Dario & Boris et al didn't have to write CC the way they did. They chose to play with and push a concept which reduced human agency -- in part because Dario concretely believes it's just "inevitable" that we should be put of of work. And his investors no doubt love this concept too.

And just like Facebook / Instagram etc. it turns out it's an addictive flow.

It remains the case there are other ways of applying LLMs and generative coding models. This modality is not intrinsic to the technology. It's being deployed this way. And humans have agency in how it's applied, even if it's hard sometimes for us to exercise it.

[SecretDreams]

> And humans have agency in how it's applied, even if it's hard sometimes for us to exercise it.

It needs to come top down from CEOs and governing bodies via regulation if we want improvements. We can't rely on the individual to not use the big red button that says "do this with no effort". We're on course for a WALL-E future if we're lucky or something far less great if we're not.

I appreciate your argument for human agency, but these types of systematic issues can't be solved bottom up.