Hacker News Reader

If the LLM is better than me at writing Torch code, it is a great idea for me to use an LLM to write my model definition, since the exact syntax or the reshaping of the tensors is not so important to me. If I want to create a convnet and train it on my images, for my own usage, I don't need to bother some Torch expert to do it for me. I can do it myself, if I understand enough about convnets themselves and not enough about Torch syntax / methods. The alternative would be to study the details of Torch in a manual, and the end result would be the same: the important thing in this task is to govern the ML concepts, not the details of MLX, Keras or PyTorch.

> I don't need to bother some Torch expert to do it for me.

This is, indeed, the core of our disagreement. You seem confident it would be a bother to others to ask for help. I'm confident that there are many who would value the opportunity to collaborate with you. I feel sure that whatever analysis you're doing could benefit from the sounding board of a domain expert, and you'd both benefit from the exchange.

EDIT: to clarify, being "famous" has nothing to do with it. Each of us has worth and we all would gain by working with others.

That's a pretty big exception to the general case though if your argument is "you are a famous person and people would love to collab" - at 2am your time, exactly when you are motivated? And what if you are not a world famous programmer, what then?

To me its like saying you shouldn't play solitaire because you are a world class poker player, and there's plenty of people who want to game with you. They are orthogonal concepts - just communicating with people can be more work than just reading on your own.

It's more of the scale you do your thing. For the quick and dirty thing, I'm going to write Torch model within the hours - where would I find a collaborator who is willing to start immediately within that time?

At least the LLM is in the same time zone...

You're right, but not always. Some people work differently than others and would rather headbutt against a wall on their own for hours than work with other people.

In the long term, it is beneficial to have experts as your collaborators. From my experience though, true collaboration is unlocked once you have established a personal relationship with someone, which takes time and repeated effort. Until then, the collaboration is no better than searching the internet or asking chatGPT.

Establishing relationships with people is hard and takes a lot of work, and frequently doesn't work out like you hope. ChatGPT is a close enough approximation for smaller tasks like the OP describes

It's hard, and many of us (myself included) and not great at it. That makes it seem easier to reach for the simulacrum. But at what cost?

Homo sapiens's superpower is social cooperation. My concern is that these systems will abet the existing social forces which seem to be causing unprecendented levels of isolation of adults, which will continue to drive smart people away from collaboration and towards solitude, at a level far beyond simple prefences would suggest.

We already have enough trouble hearing each other through the noise, and understanding what each other has to say. I don't have the answers but I'm looking for them and I do hope other humans will, too.

I think people tend to cooperate only when there are tangible benefits such as:

- survival

- making money

- sex

- enjoyment via social interactions (like parties, hangouts, etc)

It just so happens that for the majority of our civilization, to get those things, we've had to cooperate, but as we develop technology, our ability to get those things increases and our reliance on others decrease (though in a weird way it increases since technology is complexity so society becomes larger, more complex, and more inter-dependent)

We are still in an unprecedented technological boom of computing so we are adjusting on the fly to it. Like the OP says, AI can greatly accelerate independent learning, but eventually that learning plateaus. Once it does, we have to go back to collaboration, but until we find that limit, I think it's human nature to push on.

no - predatory behavior in groups is completely missing from this list. To extend that ides, IMHO much of "business" and some government, falls into this category easily. This is true for all large language groups world wide, irrespective of politics details.

Those fall under making money and surival (and honestly sex and enjoyment as well)

You said yourself. They are "social forces". In other words, problems that people created. Not technology problems.

I think there is an incorrect worldview that tries to blame human problems on technology.

It's quite true that isolation is an increasing problem. But the idea that instead of using an AI that can spit out a comprehensive answer in seconds, we should all pretend that such tools don't exist, and start constantly asking for help with every idea or request instead, while waiting 3-100 times longer for a less thorough response, is ludicrous.

It's a great idea to collaborate more and try to avoid isolation. But those are societal problems. They are not caused by the latest tools.

Also, as far as humanity's "super-power" as being collaboration, this is quite a shortsighted comment. I believe that well before AI achieves "super" level IQ, it will vastly outperform humans due to other advantages. One of those advantages is speed. Another is the ability to communicate and collaborate much, much more rapidly and effectively than humans.

One type of digital life that may take over control of the planet (possibly within decades rather than centuries) would be a type of swarm intelligence with the ability to actually "rsync" mental models to directly transfer knowledge.

Local inference of LLMs is essentially free. There doesn’t exist a sufficiently deep pool of experts of all knowledge spaces who are also willing to work for free, who can be trivially identified, contacted, and scheduled.

The argument here seems to be that there is a free supply of software developers available for the taking. Software developers are quite well paid, which suggests this is not true.

Hi Couchand.

I'm a mediocre programmer who uses GPT for a ton.

Are you volunteering to answer my questions on all the obscure stuff I ask it? Because I don't really know anybody else who will.

Anyway, my email is in my profile, write to me if this is something you're up for!

Edit: Here's a list of the stuff I asked it over the weekend:

- Discuss the pros and cons of using Standardized-audio-context instead of just relying on browser defaults. Consider build size and other issues.

- How to get github actions to cache node_modules (not the npm cache built-in to the actions/setup-node action.

- Howto get the current git hash in a GitHub action?

- Rewrite a react class-based component in functional style

- How to test that certain JSX elements were generated without directly-comparing React elements for my ANSI color to HTML parser?

- Does it make more sense to keep a copy of the original text in an editor, or just hold on to something like a crc32 to mark a document dirty?

- Can you set focus to a window you create with window.open? (You sure can!)

- Rewrite the rollup.config.js for a library of mine to produce a separate rollup config per audio worklet bundle

- Turn this tutorial for backing up a Mastodon instance into a script

- Refactor a standalone class to split it in half so each class manages precisely one thing.

- I have some code I'm writing for turn-by-turn directions. I have the data structures already, let's write code to narrate them.

- What's with this weird type error around custom CSS properties with React?

If only there was a website that allowed people to ask an absurd amount of questions, for free, and get people's insight within a semi-reasonable amount of time.

If the pytorch piece is for something low priority, its probably not worth reaching out to someone else.

I'm not going to ask a human - even a coworker - for every intellisense suggestion. Neither of us would get value from that exchange.

For small tasks this is fine, but for larger projects you have to be careful. The key to being productive with an LLM for coding is to be able to understand the code being generated, to avoid rather nasty bugs that may crop up if the model hallucinates. The worst part is that an LLM will create these bugs in very subtle ways, since it excels at writing convincing code (whether or not it actually works).

The funny thing about LLMs is that there's no rush in adopting them. They're semi useful but not really right now, but you're not gonna be 'left behind' if you don't make use of them. Everyone involved is working their hardest to make them more capable so when that day comes, you'll just use it to prompt what you want. But there's no rush to try to squeeze out anything out of the current generation which mostly lowers productivity rather than increases it.

My thinking exactly! There's FOMO going on (and being fueled by people most of which seem to hope to somehow make money off it), but the barriers to entry for using LLMs are just not that high. When the tools are good enough, I'll happily use them. Today, for the work I do, I found that not to be the case yet. I wouldn't advocate for not trying them, but I see no reason to force yourself to use them.

Yea. TBH the tooling is the bigger issue for me, personally. I come across a ton of times where an LLM might work, or could - potentially. Usually refactors. But it requires access to basically all my files, both for context and to find where references to that class/struct/etc are.

Furthermore, i'd vastly prefer a workflow most of the time where i don't even have to ask. Or so i imagine. Ie i think i'd prefer a Clippy style "Want me to write a test for this? Want me to write some documentation for this?" etc helpers. I don't want to have to ask, i want it to intuit my needs - just like any programmer could if pair programming with you.

And most of all i want it to have access to all files. To know everything about the code possible. Don't just look at the func name in isolation, attempt to understand how it's used in the project.

If i have to baby sit an LLM for a simple function refactor to give it all files where the function is used or w/e, i'd rather do it myself with tools like AST Grep or even my LSP in many cases.

I'm very interested in LLMs for simple tasks today, but the tooling feels like my primary blocker. Also possibly context length, but i think there's lots of ways around that.

Just to throw my 2c here, since I also want the models to access the whole codebase of (at least) the current project.

I had great impression of sourcegraph’s cody. https://sourcegraph.com/cody few months ago.At least with the enterprise version of the sourcegraph that had indexed most of the orgs private repos.

The web ui (vscode extension was somehow worse, not sure why) was providing damn good responses, be it code generating or QA/explanation about code spanning through multiple repos. (e.g. terraform modules living in different repos)

Afaiu it was using the sourcegraph index under the hood. But I never really deep dived into the cody’s design internals (not even sure if they are actually public)

That being said, I’ve departed from the org months ago and haven’t used cody since then, so take this with a grain of salt, since the whole comment could be outdated a lot.

I think this is an extremely unfavorable interpretation of the article. I wonder if we even read the same thing?

He sees a new tool that others have found interesting, and he identifies ways to use that tool that are useful for him, while also acknowledging where it's not useful. He backs it up with plenty of examples of where he found it not-useless. This is not a revolutionary insight, especially for a developer. We constantly use a variety of tools, such as programming languages, that have strengths and weaknesses. Why are LLMs so different? It seems foolish to claim they have zero strengths.

You might be surprised at the number of large companies who think that "GenAI" can be used to replace programmers due to non-technical executives having got the impression that a competency of LLMs is writing code ...

Of course they do have uses, but more related to discovery of APIs and documentation than actually writing code (esp. where bugs matter) for the most part.

I also have to wonder how long until open source code (e.g. GPL'd) regurgitated by LLMs and incorporated into corporate code bases becomes an issue. The C suite dudes seem concerned about employees using LLMs that may be publicly exposing their own company's code base, but illogically unworried about the reverse happening - maybe just due to not fully understanding the tech.

So you're suggesting that instead of asking an LLM, we should spend time on Fiverr/Upwork to find someone to do random coding tasks that might not fall under our expertise? Can you do that for less than 20$/month?

I agree there are difficult coordination problems that our society has failed to grapple with, let alone solve.

I don't understand which part was devastating.

A respectable person offering nuance can be shattering if you identify as pro- or anti something.

I wish we had an easier time talking about ideas with a little more detachment.

In a frictionless market it would make sense to do that. But as Coase pointed out nearly a century ago[1] forming and monitoring the relationships that allow specialization involves a certain amount of overhead. At a certain point going through the hiring or vetting process to utilize another person's skills makes sense, but it looks like the author is very far from that point.

[1]https://en.wikipedia.org/wiki/The_Nature_of_the_Firm

Bad metaphor and worse that you use it to inform your conclusion. If you must have one then use training wheels, experts don’t need training wheels beginners do, simple as that. Though the utility of LLMs goes much further as the author points out, it can often make the boring or tedious parts easier so you can add assistive peddling to the metaphor. To carry it to the end, once you have four wheels and a motor it’s not long before someone invents the car.

Like… hiring people? I think it’s a little rediculous to say “no don’t use that screwdriver, go hire a workman instead.” To say the least, there are some sizeable economic differences between those two options

> He can lucidly see that this supposed tool is completely useless to him within his area of expertise.

Did you read the article? Throughout the entire post he clearly says LLMs have a lot of value in his workflow.

Parent seems to think that Antirez has been begrudgingly pulled along by the tide of LLM hype, not as if Antirez is an accomplished developer whose judgement on tools can be trusted.

This seems like a very impractical perspective to me. If there really was a "hardware store" that we could head off to to get what we need, it might be different. But in general, that's not the case.

There can also be significant overhead in looking elsewhere for a solution. That's a big part of why so many developers reinvent things. This is often dismissed as NIH syndrome, but there's more to it than that.

You raised "introductory macroeconomics". The economic effect that will most strongly apply in the case of LLMs is that of the technology treadmill (Cochrane 1958): when there's a tool that can improve productivity, it will be used competitively so that those who don't use it effectively, to improve their productivity, will be outcompeted.

This seems like an unavoidable result in typical capitalist economies.

Your point about leveraging hungry humans would require strong incentives to overcome the treadmill effect. Most Western countries don't have many ways to implement anything like that. The closest thing might be unions, but of course most software development is not unionized.