Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI is turning us into glue

lincoln.swaine-moore.is

Post reply on HN
Print view View original

41–50 of 60 posts

Re: AI is turning us into glue

#41

Post by skydhash »

stavros wrote: post #30

So there's a wrong way and a right way to code with LLMs. The wrong way is to ask the LLM to write a bunch of code you don't understand, and to keep asking it to write more and more code to fix the problems each iteration has. That will lead to a massive tower built on sand, where everything is brittle and collapses at the slightest gust of wind. The right way is to have it autocomplete a few lines at a time for you.…

The best completion are those that let you avoiding mistyping variable names or figure out some dependency (automatically import the modules, restrict to the current scope/structure). Those has been solved for decades now. And you can get a dumb ones by doing a list of all symbols in the project directory, removing common keyword and punctuation, and do some kind of matching for filtering. The other side of the spectrum is the kind of code indexing IDEA and LSP server do.

Then you got into code boilerplate, and if you find yourself doing this a lot, that's a signal to start refactoring, add some snippets to your editor (error handling in go), write some code generators, or lament the fact that your language can't do metaprogramming.

> but I'm really not lamenting the fact that I don't have to be looking up parameters or exact function names any more.

That's a reckless attitude to have, especially if the function have drastic behavior switch like mutating argument or returning a fresh copy. All you do is assume that it behaves in certain way while the docs you haven't read will have the related warning label.

Re: AI is turning us into glue

#42

Post by skydhash »

stavros wrote: post #40
eacapeisfutuile wrote: post #37

Earlier quoted context omitted.

That level of autocomplete has been around for many years before LLM for pretty much any used language

Well, I guess I must have missed it in my thirty years of development.

Use IDEA for a java project and your TAB key will wear itself out.

Re: AI is turning us into glue

#43

Post by hnhn34 »

abdullahkhalids wrote: post #22
sarreph wrote: post #14

Earlier quoted context omitted.

On your second point - I don’t agree that humans in general will plateau. I think instead the _gap_ between humans who crave to create and learn, and those who are ostensibly potatoes, will be magnified. I see it a bit like the creator economy, where you have these maker vs consumer tranches of people.

Humans are fundamentally creators of tools and art and imaginary world - this is one of the factors that distinguishes us from animals. For most of human history, most humans spent a lot of their time creating. The very recent phenomena of a large fraction of humans creating almost nothing in their adult life is caused by modern economic systems, almost all of which force people to work so much every day that all the…

Throughout all stages of my schooling from kindergarten to college, there were always artists, tinkerers and builders, but they were never the majority.

I can't think of any time in history, under any economic system, where humans who create tools and art were the majority. Most people just want to have families and enjoy life.

Re: AI is turning us into glue

#44

Post by Havoc »

> I don't see a future where a lot of jobs don't cease to exist.

And the complete lack of a game plan of a societal level is starting to get worrying.

If we’re going to UBI this then we’re going to need a bit more of a plan than some toy studies

Re: AI is turning us into glue

#45

Post by Havoc »

rglover wrote: post #9

Really enjoyed this post. > Putting aside existential risks, I don't see a future where a lot of jobs don't cease to exist. I'm personally betting on the plateau effect with LLMs. There are two plateaus I see coming that will require humans to fix no matter what we do: 1. The LLMs themselves plateau. We're already seeing new models get worse, not better at writing code (e.g., Sonnet 3.5 seems to be better than 3.7 at…

> eventually the inputs to the LLM become stale.

Seems plausible to me that they could just keep writing python 3.13 till the end of time.

If you take say assembly - we didn’t stop writing it because it stopped working.

As a functional building block programming seems feature complete

Re: AI is turning us into glue

#46

Post by neom »

Havoc wrote: post #45
rglover wrote: post #9

Really enjoyed this post. > Putting aside existential risks, I don't see a future where a lot of jobs don't cease to exist. I'm personally betting on the plateau effect with LLMs. There are two plateaus I see coming that will require humans to fix no matter what we do: 1. The LLMs themselves plateau. We're already seeing new models get worse, not better at writing code (e.g., Sonnet 3.5 seems to be better than 3.7 at…

> eventually the inputs to the LLM become stale. Seems plausible to me that they could just keep writing python 3.13 till the end of time. If you take say assembly - we didn’t stop writing it because it stopped working. As a functional building block programming seems feature complete

"As a functional building block programming seems feature complete"

This might be one of the more fascinating things I've read in a long time. Care to expand upon? Would be genuinely curious.

Re: AI is turning us into glue

#47

Post by eximius »

I'm still pretty pessimistic on all this. Just today, I had what should have been an obvious win for an LLM coding assistant to help me. I was writing a go function that converts one very long struct into a second very long struct. The transformation was almost entirely wrapping the fields of the first struct in a wrapper in a completely rote way. If FieldA was an int on src, I wanted a dest{ FieldA: Wrapper{ Value: src.FieldA, Ratio: src.FieldA/Constant }, ... }.

It couldn't do it. I prefilled in all the fields (hundreds) and told it just to populate them, but it tried to hallucinate new fields, it would do one or two then both delete the fields I had added and add a comment saying 'then do the rest'. I tried a bunch of different prompts.

I can see how some vibe coders could make useful things, but most of my attempts to use LLMs in anything not-from-scratch are exercises in frustration.

Re: AI is turning us into glue

#48

Post by LoganDark »

blogabegonija wrote: post #25
LoganDark wrote: post #5

Earlier quoted context omitted.

I would highly caution against recommending DMT to random strangers. It is not for the faint of heart and it is also nowhere near a magic fix-all. Also, its routes of administration mostly suck (smoking/vaping or MAOIs).

Lammers should try anyway. Abstract Artificial intelligence (AI) and psychedelic medicines are among the most high-profile evolving disruptive innovations within mental healthcare in recent years. Although AI and psychedelics may not have historically shared any common ground, there exists the potential for these subjects to combine in generating innovative mental health treatment approaches https://nyaspubs.onlineli…

That application of psychedelics is almost entirely dissimilar to what you are proposing. While a lot of the emerging research in the field is nice (and long overdue in my opinion), I would be absolutely floored if any of it recommended DMT for the treatment of dissenting opinions.

It's not like I just also could use some DMT either, since I already use psychedelics on a damn near weekly basis. It's just not something that you should be telling random strangers to try. It works differently on everyone, is not always helpful, and can even be harmful.

Re: AI is turning us into glue

#50

Post by blogabegonija »

LoganDark wrote: post #48
blogabegonija wrote: post #25

Earlier quoted context omitted.

Lammers should try anyway. Abstract Artificial intelligence (AI) and psychedelic medicines are among the most high-profile evolving disruptive innovations within mental healthcare in recent years. Although AI and psychedelics may not have historically shared any common ground, there exists the potential for these subjects to combine in generating innovative mental health treatment approaches https://nyaspubs.onlineli…

That application of psychedelics is almost entirely dissimilar to what you are proposing. While a lot of the emerging research in the field is nice (and long overdue in my opinion), I would be absolutely floored if any of it recommended DMT for the treatment of dissenting opinions. It's not like I just also could use some DMT either, since I already use psychedelics on a damn near weekly basis. It's just not somethin…

Well, this field would not advance without experiments either https://scitechdaily.com/scientists-flip-two-atoms-in-lsd-an... https://www.amazon.com/AI-DMT-Simulating-Experience-through/...
Post reply on HN