"AI" is a failed technology, in the sense that it's neither fit for purpose, nor a likely path to its stated end goal of "AGI", but a lot of people's profits currently depend on pretending it's not.
-
@FreakyFwoof @skavish @Gargron
I agree with this. I oppose irresponsible use of AI, and there's a lot of that around, but I think it's also possible to use it responsibly. Responsible use gets overshadowed by the "use more AI" people that are way too common in many corporations that love mediocrity, but that doesn't mean it's impossible.
I use it myself on a project that would otherwise be way too large for me, and I think I've found an effective way to use it and get it to produce well-structured code. That does require constant code review, and clearly that code review is not perfect, because every model upgrade results in new issues being found that make me wonder how it ever missed those before. But on the whole, the end result is pretty good, and far better than the mediocre crap I often run into at big corporations. It's not as if AI is unique in bringing poorly written code to corporate projects.
But to get good code out of an LLM requires good software engineers who deeply understand how code should be structured and how to use agents to avoid the many pitfalls.
I have no doubt the future holds enormous amounts of slop.
So, pro or con? I use it myself, but I have more sympathy for anti-AI people than for "use AI everywhere" people. Am I overconfident in my ability to produce quality code with AI? Possibly. I remain critical, but will continue to use it as long as it seems to work for me.
My advice if you're not sure: don't let the AI write your code, but let it review it. It's much more thorough at review than most people are (who just skim over it and approve it). Make it an extra layer of defense, rather than a replacement.
@mcv @Gargron @skavish @fullywoolly
I'm blind so my particular use-case differs from the norm.
I see the use when people don't have the care (or time) to add alt-text and I want to be a part of the conversation.
I see uses for it at describing my surroundings.
I see uses for it when I'm in a hotel room and it's 6 in the morning, there are 3 *identical* bottles attached to the bathroom wall and I don't know which is shampoo, which is shower jell and which is conditioner because god forbid anyone bothers having a bit of difference on bottles, or even adding braille...
so yes. My use is different than most, perhaps.I also live in a world of very, very inaccessible software and web pages, despite the fat we have better standards than ever and the year is 2026.
This should be improving not disproving but sadly that is not the case.
As a result, many blind people (including myself) have decided that 'if people are going to leave us in the dark, make their portals/apps/websites inaccessible, we'll fix it ourselves.'Blind people are possibly some of the best software testers you'll ever have.
When something doesn't work, we'll find it.
You can't reach that one important app with the tab key? Stop, fix it.
Your menu bar doesn't work? Sort it out.
Alt+F for File clashes with Find for some bizarre reason? Nope.Many of us have either through necessity or sometimes lack of choice, become AI coders.
I already see some of you scoffing in the comments, and that's your right, but it's also my right to have a usable, accessible world to live in just as you do.
I fully intend to get that with, or without human help. -
@Gargron I like to ask pro-AI folks, "Where is your closest data center?" "How many acres of land did it use?" and "How much electricity will it consume?"
Many will not know the answers.
Here is a drone fly-over of a data center that's currently under construction and about 30 minutes from me. It answers many of the details everyone should be familiar with, whether you live by one, or simply utilize the technology.
https://youtu.be/52x2qBh4vsM?si=-9w4JBOBP3h1wlgO -
@Gargron
I think it's completely disingenuous to blame ai for all the troubles in the world, especially since all these problems were clearly there before ai.You have a problem with capitalism, not ai. And I do too, don't get me wrong. I'd love to live in a better, cleaner, juster world.
But ai has nothing to do with it.In fact, I think AI requires the end of capitalism. The big SciFi promise of AI is that AI will take care of all the stupid, dangerous, annoying and boring work while we play games and make art. See The Culture novels. But that's never going to happen as long as AI is controlled by powerful profit-seeking corporations.
If AI can indeed replace workers, as they claim (of which I'm very skeptical), then everybody will be poor and at their mercy, unless we completely change our economic system. (Which we should be doing anyway.)
-
@Gargron That's a quite strong stance, that missed proper argumentation. I agree it's not perfect and it has risks, but 'failed"?What is your definition for 'failed'?
Is it the fact that radiologists now can better diagnose breast cancer using AI as a tool to diagnose than without?
Or Is it the fact that it immensely progressed mRNA technology?
Or perhaps is it your fear for a 'new technology' that's alien and could perhaps take over your job and eventually your life?
We need to be careful not to equate all forms of AI. Nobody is using LLMs to diagnose breast cancer (or at least I hope not). AI has proven to be extremely effective within limited fields with clear, measurable goals.
Just like people have tried to get LLMs to play chess, and they're hilariously bad at it. But Stockfish is brilliant at it.
The big problem is the overuse of LLMs. People think that because you can have what looks like a conversation with them, they surely can do anything. They can't. They produce plausible responses, and they do that very well, which makes them extremely convincing. They're the ultimate bullshitters, and that's the big pitfall that over-eager CEOs in bullshit-sensitive industries keep falling into.
It doesn't make all AI useless, but it does make a lot of uses of AI irresponsible and dangerous.
-
@mastodonmigration cocaine is a substance with well-documented effects. AI is a tool whose results depend heavily on how it's used. the analogy doesn't work.
At this point, the cognitive decline cause by use of LLMs is also well documented and reproduced by multiple studies, so I think the analogy holds quite well.
-
calling code that’s currently being generated by AI agents “spaghetti code” is a pretty sure sign that you either haven’t actually used it seriously or don’t know how to use it.
AI-generated code can absolutely be terrible. so can human-generated code. the interesting question is whether you know how to get the agent to produce good code and how to review, test, and evolve it.
-
if you’ve never used it, then yes, you’re not in a position to judge what it looks like when used properly. being an experienced engineer and maintainer gives you plenty of expertise in software engineering, but it doesn’t magically give you first-hand experience with a tool you explicitly said you’ve never used.
you can absolutely argue that you don’t want to use it. that’s a different claim.
@skavish @Gargron Define "properly." Because I can already write code "properly" without destroying the environment with harmful data centres, shoving endless profits into middle-man companies who don't need to be involved and propping up the egos of the deranged psychopaths who are pushing this flawed technology.
-
I am pretty sure @Gargron could get great results if he wanted to use these tools. I completely respect his choice not to. but "I don't want to use it" and "it's a bad tool" are two very different things.
-
jesus fucking christ, you are arguing in circles. if you have seen good prs, then great, the tools can produce good code. if you have only seen bad ones, you know they can produce bad code. neither tells you what happens when someone who actually knows how to use the tools uses them.
this is pretty basic logic. I think I am done here
@skavish @clayote You're arguing with software engineers who don't see the point in being prompt engineers. Learning how to get a random number generator to generate the right numbers and produce the code I want is not what I've spent the last FORTY YEARS doing. It's development with extra steps, and whether you agree or not the jury is still out on the long-term benefits of your RNG.
-
@TeflonTrout please go fuck yourself properly
@skavish @TeflonTrout Insult aside, which part of his argument is wrong? Cause you sound like you're shilling for fascists.
-
I am pretty sure @Gargron could get great results if he wanted to use these tools. I completely respect his choice not to. but "I don't want to use it" and "it's a bad tool" are two very different things.
-
@mastodonmigration cocaine is a substance with well-documented effects. AI is a tool whose results depend heavily on how it's used. the analogy doesn't work.
@skavish @mastodonmigration may be good to review: https://www.frank.computer/blog/2025/05/just-a-tool.html
-
@mellowfish @Gargron as bad as it may sound, that is the unfortunate reality. People are doing absolutely mind-blowing stuff with AI that up until a few years ago was considered completely impossible from the perspective of any software engineer, yet here we are mixing videogames. AI in coding is not going away also because most companies have very generic workflows and don't really care of shipping good code if AI can manage it and work 24/7 without getting tired
@super_user_do this looks like garbage. Do you actually play and enjoy games ever??
-
@FreakyFwoof @skavish @Gargron While I agree with your ethos of neutrality here, the reality is most people who present a 'neutral' opinion are just concealing a pro or anti opinion, in your case after checking your profile, you have a recent long post praising the AI agents and data centers you are using. So ostensibly that *is* a pro opinion, you are promoting AI to your followers. Using it 'quietly' would be doing exactly that: not posting about it, nor shushing those who take anti stances.
@Shes_Gabrielle @FreakyFwoof @skavish @Gargron excellent definition of centrism, and I'll stop here. Many implications.
-
This is a cycle that AI research has gone through ever since it started: someone has an interesting idea, people are enthusiastic, first results look promising, disproportionate expectations follow, the idea leads to successful new technology that doesn't come close to the inflated expectations, and interest collapses.
AI has always been extremely hype sensitive. That doesn't mean it has no content or value, but it's always a good idea to be skeptical of those inflated expectations.
The difference is that this time, the research involves overfunded companies that believe their own hype and lots of way too rich people believe them too.
-
if you’ve never used it, then yes, you’re not in a position to judge what it looks like when used properly. being an experienced engineer and maintainer gives you plenty of expertise in software engineering, but it doesn’t magically give you first-hand experience with a tool you explicitly said you’ve never used.
you can absolutely argue that you don’t want to use it. that’s a different claim.
-
if you’ve never used it, then yes, you’re not in a position to judge what it looks like when used properly. being an experienced engineer and maintainer gives you plenty of expertise in software engineering, but it doesn’t magically give you first-hand experience with a tool you explicitly said you’ve never used.
you can absolutely argue that you don’t want to use it. that’s a different claim.
@skavish @Gargron I've used it. I got it to write part of a test. It did this 'correctly' but used an unnecessarily boilerplate-heavy approach that resulted in ten times as much code as the more concise method.
I don't use it personally at all for the past year but I know many people who do and end up reviewing dozens of LLM-assisted MRs per week, also client projects I work on in at least one case have LLMs reviewing every commit to a PR. Even the people using it successfully still allow LLM-artifacts (e.g. wrongness I know they would not introduce writing the code themselves) sometimes. I'm also less than convinced they save any time in aggregate. All the other uses of it are a very clear net loss in every context imaginable but especially my time having to read verbose crap.
-
calling code that’s currently being generated by AI agents “spaghetti code” is a pretty sure sign that you either haven’t actually used it seriously or don’t know how to use it.
AI-generated code can absolutely be terrible. so can human-generated code. the interesting question is whether you know how to get the agent to produce good code and how to review, test, and evolve it.
Human "generated" code, your phrasing of that tells so much about your stance. And also about what you think of the process, if it's just about the code as some kind of artifact, yeah you can use an LLM I guess, but code isn't just an artifact, it's a ton of experience and ways to handle common and rare cases encoded into something, I don't think a system who don't have a 100% understanding of the usecases, the ramifications even has the posibility to be what it's hyped up to.
-
ai does not have to be perfect to be useful. repeatability comes from tests, reviews, and engineering discipline. you are describing a tool that needs to be managed and pretending that proves it is useless.
You made me curious I admit. If you found out how to make agentic engineering not a waste of time... Then you probably...
1) Know how constrained agents and codebases have to be
2) Have a sense that 95% of what's critical in the job is still human
3) Invested a lot of time to find that out because it's not advertised muchYet your advice is "try and you'll see, AI can code*"
*If managed well.
🥸🥸🥸
That's a tricky statement. AI industry lives on AGI lies which produces despicable CLI tools and agentic IDE. Knowledge about how to use those things well is fragmented and contradictory. Default mode of any agent is shit.
I get why tech people are angry. Just using the tools isn't enough.
Or maybe you found some secret to make things smoother?
-
@knowprose "not all men"-ass take
@mayadev no idea what you're talking about, or why you did.
