[ClikeX]

The thing with calculator argument that always gets me. I do math unconsciously all the time. Even something as simple as adjusting a recipe when cooking. I don't need to grab a device to do that, I just do the quick math in my head. And the only way I can do that is because I learned how to do math. And even with a calculator, I still needed to know what to calculate. The argument "we don't need math, we have a calculator" assumes you always get a textbook question that lays it out for you.

Same goes for LLMs. I can use them for programming, and they're very convenient. But I still need to know what to ask it and make sure it stays within the confines of what I want. And without my knowledge I would have no clue if what it's trying do is correct, or safe.

Naturally, this assumes a workflow where you do actually look and modify the output yourself. But I'd argue that any non tech person is inevitably going to hit a wall where they can't debug themselves out of without getting a human involved.

[delusional]

> I just do the quick math in my head. And the only way I can do that is because I learned how to do math.

A lot of Hacker news commenters tend to overestimate human ability without education. That is, the tend to believe that people are able to do a lot more without regular training than they actually are. They believe that "math" is some naturally ordained eventuality that humans just do, or likewise that "reasoning" is some immutable natural behavior. In truth these things are "unnatural" human reinforced structures that we have to learn and adapt our brains into.

[Rendello]

I'm abhorrent at arithmetic and I think it affects me greatly. I know quite a few people who are barely literate and manage to get by in life, but it obviously makes things hard for them and strips them of opportunities (and they forego holding any identity that seems to "smart" for themselves). I think numeracy is similar.

I wonder how much better I could become as an adult at mental math. I did a math course a few years ago (Math Academy) and put some effort into improving my addition, subtraction, division, and multiplication, but I wish I'd done more up front. Even one-figure arithmetic is so fuzzy to me, like it's a hashtable where the keys don't line up to the right values. Can I unlearn that now? O(1) doesn't matter if the result is wrong.

[abalashov]

Underrated and paradoxical element of this, to which, as far as I can tell, LLM boosters offer no solution:

> And without my knowledge I would have no clue if what it's trying do is correct, or safe.

I would contend you got the knowledge by typing the code yourself, that there's no other way to get it, and that if you stop typing the code yourself--and the slogging that entails--you'll lose the ability to prompt LLMs effectively.

It's not that I think the physiokinetic aspects of typing as an input mechanism hold some metaphysical distinction, but rather the level of engagement it forces with the code, and the units in which it does so. I'm not aware as yet of any viable replacement for that.

It's easy to trade on decades of software engineering experience with LLMs: with sufficient experience, everything goes around and comes around, almost any pattern is recognisable, the gratification is immediate, the benefits are now, while the costs and disasters are down the road.

However, the technology world is not static, and if you don't keep up with new frameworks, libraries, languages and other tech in that physical-mechanical "mind-body-keyboard" way that typing--or something substantially close to it--accomplishes, you will lose the ability to navigate that world fluently. To say it's just another abstraction layer and that the world didn't crumble after compilers is to miss something quite essential about how LLMs differ from compilers or high-level languages. The disengagement with the process of physically programming something quite specific will take down with it the ability, over time, to formulate useful prompts and competently review the output.

[ClikeX]

> I would contend you got the knowledge by typing the code yourself, that there's no other way to get it, and that if you stop typing the code yourself--and the slogging that entails--you'll lose the ability to prompt LLMs effectively.

For sure, and I've tried picking up a new language with the use of LLMs, but the concepts just don't stick because I don't actually do the work. That's why I do try to limit my use of LLMs to fields I'm already closely familiar with, and also keep its output contained to actually reviewable chunks. Or things that are just tedious, like OCR and large text transformations.

[abalashov]

> the concepts just don't stick because I don't actually do the work

Yes! And this is the part of the de-skilling puzzle that is completely unaddressed by AI boosters.

Maybe LLMs are a force multiplier, but there still has to be some force to multiply, and I don't think a lot of folks ask the question of how that force is actually cultivated. This nebulous, airy-fairy notion that humans add "architecture" or "taste" doesn't tell the story of how, concretely, they came to have it. It seems to me there is no escaping that it came from typing the code.

Like you, I have a much better grasp of code and API surfaces I've physically typed than things LLMs have emitted and I have reviewed, in a conceptual sense, but which I could not have typed myself then, nor can type now.

[intended]

Yes! If anything both, the supply of veterans and the ratio of judgement capacity to generated content, will only go down with time.

So where are the new veterans going to be coming from? You can’t spin up experienced workers overnight.

Essentially the same supply chain issue which nation states are worrying about today, except for skill.

[maybiiLen]

I think you are underestimating how fast this stuff is moving. Calling AI "just autocomplete" for system architecture is outdated.

If you were to ask it (especially with Claude's newer model, Fable 5) to draft an architecture design diagram, it would do a better job than any junior dev ever could. Sure, giving context is still necessary, but you can just speak in plain, normal English and it will understand. AI is getting scary good at system design.

[ClikeX]

> Calling AI "just autocomplete" for system architecture is outdated.

I didn't say that, though.

> it would do a better job than any junior dev ever could. Sure, giving context is still necessary, but you can just speak in plain, normal English and it will understand. AI is getting scary good at system design.

Sure. But that's not my point. Don't you still want those juniors to be capable of creating those diagrams? So they can understand them? If the juniors don't get what the AI is outputting, why are you even creating the diagrams in the first place, right?

That's what this topic is about. Just like how calculators helped us do complex tasks, we still needed to understand math. Same with AI, you still need to be understand the topics you ask it to do.