Hacker News Reader

Comment by ltbarcly3 | Hacker News Reader

I think anyone who thinks that LLMs are not intelligent in any sense is simply living in denial. They might not be intelligent in the same way a human is intelligent, they might make mistakes a person wouldn't make, but that's not the question.

Any standard of intelligence devised before LLMs is passed by LLMs relatively easily. They do things that 10 years ago people would have said are impossible for a computer to do.

I can run claude code on my laptop with an instruction like "fix the sound card on this laptop" and it will analyze what my current settings are, determine what might be wrong, devise tests to have me gather information it can't gather itself, run commands to probe hardware for it's capabilities, and finally offer a menu of solutions, give the commands to implement the solution, and finally test that the solution works perfectly. Can you do that?

View on HN

I'm vibe coding now, after work. I am able to much more quickly explore the landscape of a problem, get into and out of dead ends in minutes instead of wasting an evening. At some point I need to go in and fix, but the benefit of the tool is there. It is like a electric screwdriver vs. normal one. Sometimes the normal one can do things the electric can't, but hell if you get an IKEA deliver you want the electric one.

Got any recent specific examples of it saving you an entire evening?

0. Claude, have a look at frontend project A and backend project B.

1. create a skeleton clone of frontend A, named frontend B, which is meant to be the frontend for backend project B, including the oAuth configuration

2. create the kubernetes yaml and deployment.sh, it should be available under b.mydomain.com for frontend B and run it, make sure the deployment worked by checking the page on b.mydomain.com

3. in frontend B, implement the UI for controller B1 from backend B, create the necessary routing to this component and add a link to it to the main menu, there should be a page /b1 that lists the entries, /b1/xxx to display details, /b1/xxx/edit to edit an entry and /b1/new to create one

4. in frontend B, implement the UI for controller B2 from backend B, create the necessary routing to this component and add a link to it to the main menu, etc.

etc.

All of this is done in 10 minutes. Yeah I could do all of this myself, but it would take longer.

Did you need it though? Like most projects I see being done by people with Claude Code are just their personal projects, which they wouldn't have wasted their time on in the past but now they will get pulled into the terminal thinking its only gonna take 20 mins and they end up burning 100s of subscription dollars on it. If there is no other maintainer & the project is all yours, I dont see any harm in doing it.

I don't NEED this, but turning 2-3 hours fiddling with DTOs, kubernetes yaml, dockerfiles, deployment scripts and other busy work into half an hour does "save you the whole evening" (which our current discussion is about!).

With adults usually having no more than 2-3 hours of free time per work day, this allows you to productively program in your free time without fully burning out.

Also, my company pays for claude and does not give a shit what I do with it.

Bad example. IKEA assembles better with a manual screwdriver.

You wouldn't say that anymore if you would have ever assembled PAX doors.

Maybe? I'm not familiar with every ikea product. But it looks like it take a dozen small screws into soft wood.

And is the electric one intelligent? :p

Who cares!

It's incredible that on Hacker News we still encounter posts by people who will or cannot differentiate mathematics from magic.

Intelligence is not magic though. The difference between intelligence and mathematics can plausibly be the same kind of difference between chemistry and intelligence.

There is Intelligence and there is Imitation of Intelligence. LLMs do the latter.

Talk to any model about deep subjects. You ll understand what I am saying. After a while it will start going around in circles.

FFS ask it to make an original joke, and be amused..

Many animals are clearly intelligent, but I can't talk to them at all.

If you understood anything about this topic you would know that imitation of intelligence requires intelligence. It's the same thing.

Give it some thought.

It has been given significant thought already. This isn't a new topic.

> After a while it will start going around in circles.

so like your average human

> FFS ask it to make an original joke, and be amused..

let's try this one on you - say an original joke

oh, right, you dont respond to strangers prompts, thus you have agency, unlike an LLM

>so like your average human

If an average human has seen and read all that is written till now, I bet that they can hold the conversation going for quite a long time...

>say an original joke

I asked an LLM if it had a good night with sweet dreams, It said, "I don't sleep and I only dream when I work!"

Human intelligence is chemistry and biology, not magic. OK, now what?

Your brain is just math implemented in wet meat.

it's also incredible we find people which can't differentiate physics/mathematics from the magic of the human brain

There are dozens of definitions of "intelligence", we can't even agree what intelligence means in humans, never mind elsewhere. So yes, by some subset of definitions it is intelligent.

But by some subset of definitions my calculator is intelligent. By some subset of definitions a mouse is intelligent. And, more interestingly, by some subset of definitions a mouse is far more intelligent than an LLM.

It is the imitation of intelligence.

It works because people have answered similar questions a million times on the internet and the LLMs are trained on it.

So it will work for a while. When the human generated stuff stops appearing online, then LLMs ll quickly fall in usefulness.

But that is enough time for the people who might think that it going to last for ever to make huge investments into it, and the AI companies to get away with the loot.

Actually it is the best kind of scam...

EDIT: Another thought. Thus it seems that AI companies actually have an incentive to hinder developements, because new things mean that their model is less useful. With the widespread dependence on AI, they might even get away with manipulating the population to stagnate.

I also cannot calculate the square root of 472629462.

My pocket calculator is not intelligent. Nor are LLMs.

You'd be surprised. You could probably get three digits of the square root in under a minute if you tried.

I did that when I was 14 because I had no other choice, damn you SoundBlaster! I didn't get any menu but I got sound in the end.

I don't think conflating intelligence with "what a computer can do" makes much sense though. I can't calculate the X digit of PI in less than Z, I'm still intelligent (or I pretend to be).

But the question is not about intelligence, it's a red herring, it's just about utility and they (LLM's) are useful.

set BLASTER=A220 I5 D1

>I can run claude code on my laptop with an instruction like "fix the sound card on this laptop" and it will analyze what my current settings are, determine what might be wrong, devise tests to have me gather information it can't gather itself, run commands to probe hardware for it's capabilities, and finally offer a menu of solutions, give the commands to implement the solution, and finally test that the solution works perfectly. Can you do that?

Yes, I have worked in small enough companies in which the developers just end up becoming the default IT help desk. I never had any formal training in IT, but most of that kind of IT work can be accomplished with decent enough Google skills. In a way, it worked the same as you and the LLM. I would go poking through settings, run tests to gather info, run commands, and overall just keep trying different solutions until either one worked or it became reasonable to give up. I'm sure many people here have had similar experiences doing the same thing in their own families. I'm not too impressed with an LLM doing that. In this example, it's functionally just improving people's Googling skills.

Everything is magic when you don't understand how things work.

In a way LLMs are intelligence tests indeed.

Reply on news.ycombinator.com