Live data from Hacker News

I don't care how well your "AI" works

fokus.cool

781–790 of 808 posts

Re: I don't care how well your "AI" works

#781

I honestly don't get vibe coding. I've tried it multiple times, but even after spending 4 hours on a fresh project I don't feel like I know what the hell is going on anymore. At that point I'm just guessing what the next prompt is to make it work. I have no critical knowledge about the codebase that makes me feel like I could fix an edge case without reading the source code line by line (which at that point would pro…

It's interesting that this is a similar criticism to what was levelled at Ruby on Rails back in the day. I think generating a bunch of code - whether through AI or a "framework" - always has the effect of obscuring the mental model of what's going on. Though at least with Rails there's a consistent output for a given input that can eventually be grokked.

A framework provides axioms that your theory-building can work on top of. It works until there are bugs or inconsistencies in the framework that mean you can't trust those axioms.

Re: I don't care how well your "AI" works

#782

Earlier quoted context omitted.

better integrations won't do anything to fix the fact that these tools are, by their mathematical nature, unreliable and always will be

so are people

But humans have vastly lower error rates than llms. And in a multi-step process that means that those error rates compound. And when that happens, you end up with a 50/50 or worse

Re: I don't care how well your "AI" works

#783

Earlier quoted context omitted.

>That is a failure of basic observation. A failure to see the thing you don't like because you don't like and decide not to look. Will you like it if you look? Maybe we're observing different parts of the elephant. This is my industry right now: https://www.pcgamer.com/games/call-of-duty/call-of-duty-blac... A deca-billion dollar franchise now owned by a trillion dollar tech company... using it to make art that would…

Yep. They’re just happy to light any sense of QC on fire. Such overt contempt for pretty much everyone directly involved in the economic transaction. But not the investors, of course.

most people don't care and don't even know, and Activision correctly believed that it won't matter (doesn't matter)

... of course the whole company from Satya down to the idiot whatever art director who asked for this are impervious to shame, so yes, how it could matter? it's not like people bought the last ~10 years of CoD for its artistic value, people bought it because they wanted to switch off their brain and get shot and shouted at by teenagers

Re: I don't care how well your "AI" works

#784
post #83
post #23

And then there is the moderate position: Don't be the person refusing the use a calculator / PC / mobile phone / AI. Regularly give the new tool a chance and check if improvements are useful for specific tasks. And carry on with your life.

I was once in your camp, thinking there was some sort of middle-ground to be had with the emergence of Generative AI and it's potential as a useful tool to help me do more work in less time, but I suppose the folks who opposed automated industrial machinery back in the day did the same. The problem is that, historically speaking, you have two choices; 1. Resist as long as you can, risking being labeled a Luddite or w…

Maybe its just me, but i often feel that the issue in these debates is not that we give up creativity but people unwillingness to change their creativity.

When PCs came around, people looked down on the idea of painting on a screen. Some stubborn held on to their easel, while the often younger generation embraced the new tech, and made their carriers.

LLMs are the same ... We have people who are stubborn and still want to do everything on their easel and good for them. But those that adapted, will turn out more work and eventually replace most of the die hards in the workplace. Sure, there will be people who are needed, just like we had Fortran programmers 50 years later making bank, or painters who make bank.

But the idea that it makes us less creative is stupid. Did PCs make us less creative in painting? No, we got a ton of new media and changes. We adapted to the tools and possibilities.

O, PCs came on the market, well, no way somebody is going to use that to make ... MUSIC ... you only do that with real instruments. Que entire generation of techno, movie music etc all made digitally. You did not need to be a directory, know how to play dozens of different instruments to make insane pieces of music. You used your creativity to use the tools at your disposal !

The fact that you can now do the work in a few weeks, that will have taken you a year as a programmer, that opens up the doors for more creativity. Prototyping a idea does not take months, but days. It changes the industry...

Why do i need to pay license cost for a piece of software that is "enterprise", when i can now make the same level of software in a few weeks. It actually take the power away from big corporations, sure, you rely on the tool, and whoever is behind the tool for now (like anthropic etc) but as time moves forwards, more hardware will become more powerful, and LLMs will be more in the open source, ...

Remember what i said about music ... hey, maybe creative people will be able to make music that is different thanks to AI. Hey, you wanted to make a Magic The Gathering game but the art was a delima, ... LLMs suddenly open the door to make new products, and change the industry away from large corporation where the entry fee is high.

We are on a threshold of change, those that stay behind and think it kills creativity, never really used the new tools.

I am writing software right now, that will have took me a year to write. It duplicates the function of some enterprise software that will have cost me insane money. BUTTTTTT, because i now control this software, i can add features that i always found lacking or missing, because that enterprise software only looked at the companies, now what somebody like me may need.

All for the low low price of barely $40, what is probably 70~100k in developer cost. That is still being creative, i made something new using a existing idea, with my own touches to it. But i used a tool for it... Eventually, my code if open source, may be used in other LLMs to improve them. What in turn makes them smarter and maybe some of my idea get used by others.

This is frankly how we as the human race advanced, not by stagnating and not embracing change, but by often copying, improving, using our creativity. Does that mean i need to know how a LLM works? No, just like do not need to know how a bread slicer works, ...

Anyway, ... this discussion is never going to end, because there is always resistance to change. I remember my parents about smartphones, ... guess who uses them, even at their old age. Eventually some power always get consolidated but for now, with the progress i am seeing in the open source/open weights LLM progress, there will been a escape hatch for those that do not want to be linked to specific corporations, just like Linux etc exists.

Re: I don't care how well your "AI" works

#785

Earlier quoted context omitted.

I'm firmly in the “don't want to use it; if you want to, feel free, but stop nagging me to” camp. Oh, and the “I'm not accepting 'the AI did it' as an excuse for failures” camp. Just like outsourcing to other humans: you chose the tool(s), you are responsible for verifying the output. I got into programming and kicking infrastructure because I'm the sort of sad git who likes the details, and I'm not about to let some…

“Don’t want… “not accepting” Now this, putting aside my thoughts above, i find a compelling argument. You just don’t want to. I think that should go along with a reasonable understanding of what a person is choosing to not use, but I’ll presume you have that. Then? Sure, the frustrating part is to see someone making that choice tell other people that theirs is invalid, especially when we don’t know what the scene wil…

I don't disagree, but I feel quoting me like

> “Don’t want… “not accepting”

misrepresents my position a bit. Those were quite separate thoughts.

While I don't want to, I accept that others do, and will, and will be productive doing so.

What I won't accept is an attitude to bugs along the lines of “ah, that was the AI, what can you do?”, which implies either the tool is trusted far too much or people are simply being lazy. This already seems to be creeping in in places, and causes concerns about both security and stability if it is allowed to proliferate.

Re: I don't care how well your "AI" works

#786

Earlier quoted context omitted.

Well your anecdote is clearly at odds with absolutely all of the macro economic data.

Actually, it’s not. The job market changed about a year or so before ChatGPT

The idea that there is a hiring spree of developers doesn’t jibe with reality

Re: I don't care how well your "AI" works

#787

Earlier quoted context omitted.

so are people

But humans have vastly lower error rates than llms. And in a multi-step process that means that those error rates compound. And when that happens, you end up with a 50/50 or worse

And, more importantly, a given human can, and usually will, learn from their mistakes and do better in a reasonably consistent pattern.

And when humans do make mistakes, they're also in patterns that are fairly predictable and easy for other humans to understand, because we make mistakes due to a few different well-known categories of errors of thought and behavior.

LLMs, meanwhile, make mistakes simply because they happen to have randomly generated incorrect text that time. Or, to look at it another way, they get things right simply because they happen to have randomly generated correct text that time.

Individual humans can be highly reliable. Humans can consciously make tradeoffs between speed and reliability. Individual unreliable humans can become more reliable through time and effort.

None of these are true of LLMs.

Re: I don't care how well your "AI" works

#788
post #783

Earlier quoted context omitted.

Yep. They’re just happy to light any sense of QC on fire. Such overt contempt for pretty much everyone directly involved in the economic transaction. But not the investors, of course.

most people don't care and don't even know, and Activision correctly believed that it won't matter (doesn't matter) ... of course the whole company from Satya down to the idiot whatever art director who asked for this are impervious to shame, so yes, how it could matter? it's not like people bought the last ~10 years of CoD for its artistic value, people bought it because they wanted to switch off their brain and get…

>and Activision correctly believed that it won't matter

The titanic had 3 days off warning. It took 3 hours to sink.

It "won't matter" until it does. And then it will collapse all at once. WWE 2K20 shows that even yearly franchises have breaking points where the backlash is too much to ignore.

But i guess Activision executives are thinking thwy will have jumped ship if and when that time hits.

Re: I don't care how well your "AI" works

#789

Earlier quoted context omitted.

Likely progressive, but definitely not luddite [0]. Anti-capitalist for sure. I struggle with this discourse deeply. With many posters like OP, I align almost completely - unions are good, large megacorps are bad, death to facists etc. It's when we get to the AI issue that I do a bit of a double take. Right now, AI is almost completely in the hands of a few large corp entities, yes. But once upon a time, so was the i…

Not true for the Internet. It was the open system anyone could join and many people were shocked it succeeded over the proprietary networks being developed.

It started as ARPANET, which was not an open system (slightly sloppy use of the term i know, not technically the internet but I consider one the extension of the other).

Re: I don't care how well your "AI" works

#790

Earlier quoted context omitted.

Have you seen the latest AI slop in game design lately, destroying human creativity? This just in: 90% of everything is crap. AI does not, cannot, and will not change that. Have you seen how this tech is being used to control narratives to subjugate populations to the will of authoritarian governments? Can't say as I have. The only authoritarians in this thread are the ones telling us what we should and should not be…

> This just in: 90% of everything is crap. AI does not, cannot, and will not change that. Use a search engine. Such a disingenuous take. AI is demonstrably being used to flood the web with slop faster than humans ever could. Now 99% of everything will be crap, just like your takes.

Now 99% of everything will be crap, just like your takes.

And the worst part is, they made it without checking with you first. Must be frustrating.

Post reply on HN