[throwout4110]

Right this is what I can’t quite understand. A lot of HN folks appear to have been burned by e.g. horrible corporate or business ideas by non technical people that don’t understand AI, that is completely understandable. What I never understand is the population of coders that don’t see any value in coding agents or are aggressively against them, or people that deride LLMs as failing to be able to do X (or hallucinate etc) and are therefore useless and every thing is AI Slop, without recognizing that what we can do today is almost unrecognizeable from the world of 3 years ago. The progress has moved astoundingly fast and the sheer amount of capital and competition and pressure means the train is not slowing down. Predictions of “2025 is the year of coding agents” from a chorus of otherwise unpalatable CEOs was in fact absolutely true…

[aisengard]

There is zero guarantee that these tools will continue to be there. Those of us who are skeptical of the value of the tools may find them somewhat useful, but are quite wary of ripping up the workflows we've built for ourselves over decade(s)(+) in favor of something that might be 10-20% more useful, but could be taken away or charged greater fees or literally collapse in functionality at any moment, leaving us suddenly crippled. I'll keep the thing I know works, I know will always be there (because it's open source, etc), even if it means I'm slightly less productive over the next X amount of time otherwise.

[throwout4110]

What would you imagine a plausible scenario would possibly be that your tools would be taken away or “collapse in functionality”? I would say Claude right now has probably made worse code and wasted time than if I had coded things myself, but it’s because this is like the first few hundred days of this. Open weight models are also worse but they will never go away and improve steadily as well. I am all for people doing whatever works for them I just don’t get the negativity or the skepticism when you look at the progress over what has been almost zero time. It’s crappy now in many respects but it’s like saying “my car is slow” in the one millisecond after I floor the gas pedal

[watwut]

> What would you imagine a plausible scenario would possibly be that your tools would be taken away or “collapse in functionality”?

Simple. The company providing the tool needs actual earning suddenly. Therefore, they need to raise the prices. They also need users to spend more tokens, so they will make the tool respond in a way that requires more refinement. After all, the latter is exactly what happened with google search.

At this point, that is pretty normal software cycle - try to attract crowd by being free or cheap, then lock features behind paywall. Then simultaneously raise prices more and more while making the product worst.

This literally NEEDS to happen, because these companies do not have any other path to profitability. So, it will happen at some point.

[throwout4110]

Sure but you’re forgetting that competition exists. If anthropic investors suddenly say “enough” and demand positive cash flow it wouldn’t be that hard, everyone is capturing users for flywheels and capex for model improvements because if they don’t they will be guaranteed to lose.

It’s going to definitely be crappy, remember Google in 2003 with relevant results and no endless SEO , or Amazon reviews being reliable, or Uber being simple and cheap, etc. once growth phase ends monetization begins and experience declines but this is guard railed by the fact that there are many players.

[watwut]

Comsidering what I described is how tech companies actually function and functioned in the past, theoretical competition wont help.

They are competing themselves into massive unprofitability. Eventually they will die or do the above in cooperation. Maybe there will bw minor snandal about it, but that sort of collution is not prosecuted or seriously investigated if done by big companies.

So, it will happen exactly as it always happens with tech.

[takluyver]

My understanding is that all the big AI companies are currently offering services at a loss, doing the classic Silicon Valley playbook of burning investor cache to get big, and then hope to make a profit later. So any service you depend on could crash out of the race, and if one emerges as a victorious monopoly and you rely on them, they can charge you almost whatever they like.

To my mind, the 'only just started' argument is wearing off. It's software, it moves fast anyway, and all the giants of the tech world have been feverishly throwing money at AI for the last couple of years. I don't buy that we're still just at the beginning of some huge exponential improvement.

[ben_w]

My understanding is they make a loss overall due to the spending on training new models, that the API costs are profit making if considered in isolation. That said, this is based on guestimates based on hosting costs of open-weight models, owing to a lack of financial transparancey everywhere for the secret-weights models.

[blibble]

> that the API costs are profit making if considered in isolation.

no, they are currently losing money on inference too

[mjr00]

> Predictions of “2025 is the year of coding agents” from a chorus of otherwise unpalatable CEOs was in fact absolutely true…

... but maybe not in the way that these CEOs had hoped.[0]

Part of the AI fatigue is that busy, competent devs are getting swarmed with massive amounts of slop from not-very-good developers. Or product managers getting 5 paragraphs of GenAI bug reports instead of a clear and concise explanation.

I have high hopes for AI and think generative tooling is extremely useful in the right hands. But it is extremely concerning that AI is allowing some of the worst, least competent people to generate an order of magnitude more "content" with little awareness of how bad it is.

[0] https://github.com/ocaml/ocaml/pull/14369

[justatdotin]

> busy, competent devs are getting swarmed with massive amounts of slop from not-very-good developers

that is a real issue and yet a normal problem and so has an obvious response.

oh wow that PR

[bigstrat2003]

> What I never understand is the population of coders that don’t see any value in coding agents or are aggressively against them, or people that deride LLMs as failing to be able to do X (or hallucinate etc) and are therefore useless and every thing is AI Slop, without recognizing that what we can do today is almost unrecognizeable from the world of 3 years ago.

I don't recognize that because it isn't true. I try the LLMs every now and then, and they still make the same stupid hallucinations that ChatGPT did on day 1. AI hype proponents love to make claims that the tech has improved a ton, but based on my experience trying to use it those claims are completely baseless.

[ben_w]

> I try the LLMs every now and then, and they still make the same stupid hallucinations that ChatGPT did on day 1.

One of the tests I sometimes do of LLMs is a geometry puzzle:

  You're on the equator facing south. You move forward 10,000 km along the surface of the Earth. You are rotate 90° clockwise. You move another 10,000 km forward along the surface of the earth. Rotate another 90° clockwise, then move another 10,000 km forward along the surface of the Earth.

  Where are you now, and what direction are you facing?

They all used to get this wrong all the time. Now the best ones sometimes don't. (That said, only one to succed just as I write this comment was DeepSeek; the first I saw succeed was one of ChatGPT's models but that's now back to the usual error they all used to make).

Anecdotes are of course a bad way to study this kind of thing.

Unfortunately, so are the benchmarks, because the models have quickly saturated most of them, including traditional IQ tests (on the plus side, this has demonstrated that IQ tests are definitely a learnable skill, as LLMs loose 40-50 IQ points when going from public to private IQ tests) and stuff like the maths olympiad.

Right now, AFAICT the only open benchmarks are the METR time horizon metric, the ARC-AGI family of tests, and the "make me an SVG of ${…}" stuff inspired by Simon Willison's pelican on a bike.

[Smaug123]

Out of interest, was your intended answer "where you started, facing east"?

FWIW, Claude Opus 4.5 gets this right for me, assuming that is the intended answer. On request, it also gave me a Mathematica program which (after I fixed some trivial exceptions due to errors in units) informs me that using the ITRF00 datum the actual answer is 0.0177593 degrees north and 0.168379 west of where you started (about 11.7 miles away from the starting point) and your rotation is 89.98 degrees rather than 90.

(ChatGPT 5.1 Thinking, for me, get the wrong answer because it correctly gets near the South Pole and then follows a line of latitude 200 times round the South Pole for the second leg, which strikes me as a flatly incorrect interpretation of the words "move forward along the surface of the earth". Was that the "usual error they all used to make"?)

[ben_w]

> Out of interest, was your intended answer "where you started, facing east"?

Or anything close to it so long as the logic is right, yes. I care about the reasoning failure, not the small difference between the exact quarter-circumferences of these great circles and 10,000km; (Not that it really matters, but now you've said the answer, this test becomes even less reliable than it already was).

> FWIW, Claude Opus 4.5 gets this right for me, assuming that is the intended answer.

Like I said, now the best ones sometimes don't [always get it wrong].

For me yesterday, Claude (albeit Sonnet 4.5, because my testing is cheap) avoided the south pole issue, but then got the third leg wrong and ended up at the north pole. A while back ChatGPT 5 (I looked the result up) got the answer right, yesterday GPT-5-thinking-mini (auto-selected by the system) got it wrong same way as you report on the south pole but then also got the equator wrong and ended up near the north pole.

"Never" to "unreliable success" is still an improvement.

[xantronix]

Yeah, I'm pretty sure that's correct. Just whipped this up, using the WGS-84 datum.

  (use-modules (geo vincenty))
  
  (let walk ((p '(0 0 180))
             (i 0))
    (cond ((= i 3)
           (display p)
           (newline))
          (else
            (walk (apply vincenty
                         (list (car p) (cadr p) (+ 90 (caddr p)) 10000000))
                  (+ i 1)))))

Running this yields:

  (0.01777744062090717 0.16837322410251268 179.98234155229127)

Surely the discrepancy is down to spheroid vs sphere, yeah?

[hectdev]

This fascinates me. Just observing but because it hasn't worked for you, everyone else must be lying? (I'm assuming that's what you mean by baseless)

How does that bridge get built? I can provide tangible real life examples but I've found push back from that in other online conversations.

[queenkjuul]

My boss has been passing off Claude generated code and documentation to me all year. It is consistently garbage. It consistently hallucinates. I consistently have to rewrite most, if not all, of what I'm handed.

I do also try and use Claude Code for certain tasks. More often than not, i regret it, but I've started to zero in on tasks it's helpful with (configuration and debugging, not so much coding).

But it's very easy then for me to hear people saying that AI gives them so much useful code, and for me to assume that they are like my boss: not examining that code carefully, or not holding their output to particularly high standards, or aren't responsible for the maintenance and thus don't need to care. That doesn't mean they're lying, but it doesn't mean they're right.

[hectdev]

Not everyone is your boss. I have 15 years of experience coding. So when the AI hallucinates, I call that out and it improves the code it does create. If someone is passing off Ai's first pass as done, they are not using the tool correctly.

[queenkjuul]

My boss has 28 years of experience coding so that clearly isn't the deciding factor here.

Yes, i suppose it is theoretically possible that you are that much better than my boss and i at coaxing good output from an LLM, but I'm going to continue to be skeptical until i see it with my own eyes.

[int_19h]

"Claude Code" by itself is not specific enough; which model are we talking about?

[bluefirebrand]

> it hasn't worked for you, everyone else must be lying?

Well, some non-zero amount of you are probably very financially invested in AI, so lying is not out of the question

Or you simply have blinders on because of your financial investments. After all, emotional investment often follows financial investment

Or, you're just not as good as you think you are. Maybe you're talking to people who are much better at building software than you are, and they find the stuff the AI builds does not impress them, while you are not as skilled so you are impressed by it.

There are lots of reasons someone might disagree without thinking everyone else is lying

[hectdev]

I think calling it baseless to claim benefits from AI is more than disagreeing. It's claiming a rightness that is just contrarian and hyperbolic. It's really interesting to me that the skeptics are exactly who should be using AI. Push back on it. Tell it that the code it made was wrong.

[shepherdjerred]

What have you tried? How much time have you spent? Using AI is it’s own skill set separate from programming

[elictronic]

AI is in a hype bubble that will crash just like every other bubble. The underlying uses are there but just like Dot Com, Tulips, subprime mortgages, and even Sir Isaac Newton's failings with the South Sea Company the financial side will fall.

This will cause bankruptcies and huge job losses. The argument for and against AI doesn't really matter in the end, because the finances don't make a lick of sense.

[throwout4110]

Ok sure the bubble/non-bubble stuff, fine, but in terms of “things I’d like to be a part of” it’s hard to imagine a more transformative technology (not to again turn off the anti-hype crowd). But ok, say it’s 1997, you don’t like the valuations you see. But as a tech person you’re not excited by browsers, the internet, the possibilities? You don’t want to be a part of that even if it means a bubble pops? I also hear a lot of people argue “finances don’t make a lick of sense” but i don’t think things are that cut and dried and I don’t see this as obvious. I don’t think really many people know how things will evolve and what size a market correction or bubble would have.

[zdragnar]

What precisely about AI is transformative, compared to the internet? E-mail replaced so much of faxing, phoning and physical mail. Online shopping replaced going to stores and hoping they have what you want, and hoping it is in stock, and hoping it is a good price. It replaced travel agents to a significant degree and reoriented many industries. It was the vehicle that killed CDs and physical media in general.

With AI I can... generate slop. Sometimes that is helpful, but it isn't yet at the point where it's replacing anything for me aside from making google searches take a bit less time on things that I don't need a definitive answer for.

It's popping up in my music streams now and then, and I generally hate it. Mushy-mouthed fake vocals over fake instruments. It pops up online and aside from the occasional meme I hate it there too. It pops up all over blogs and emails and I profoundly hate it there, given that it encourages the actual author to silence themselves and replaces their thoughts with bland drivel.

Every single software product I use begs me to use their AI integration, and instead of "no" I'm given the option of "not now", despite me not needing it, and so I'm constantly being pestered about it by something.

It has, thus far, made nearly everything worse.

[throwout4110]

> With AI I can... generate slop. Sometimes that is helpful, but it isn't yet at the point where it's replacing anything for me aside from making google searches take a bit less time on things that I don't need a definitive answer for.

I think this is probably the disconnect, this seems so wildly different from my experience. Not only that, I’ll grant that there are a ton of limitations still but surely you’d concede that there has been an incredible amount of progress in a very short time? Like I can’t imagine someone who sits down with Claude like I do and gets up and says “this is crap and a fad and won’t go anywhere”.

As for generated content, I again agree with you and you’d be surprised to learn that _execs_ agree with you but look at models from 1, 2, 3 years ago and tell me you don’t see a frightening progression of quality. If you want to say “I’ll believe it when I see it” that’s fine but my god just look at the trajectory.

For AI slop text, once again agree, once again I think we all have to figure out how to use it, but it is great for e.g. helping me rewrite a wordy message quickly, making a paper or a doc more readable, combining my notes into something polished, etc, and it’s getting better and better and better.

So I disagree it has made everything worse but I definitely agree that it has made a lot of things worse and we have a lot of Pets.com ideas that are totally not viable today, but the point I think people are maybe missing (?) is that it’s not about where we are it’s about the velocity and the future. You may be terrified and nauseated by $1T in capex on AI infra, fine but what that tells you is the scale is going to grow even further _in addition_ to the methodological / algorithmic improvements to tackle things like continual learning, robustness, higher quality multimodal generation with e.g. true narrative consistency, etc etc etc. in 5 years I don’t think many people will think of “slop” so negatively

[zdragnar]

Where you see exponential growth in capability and value, I see the early stages of logarithmic growth.

A similar thing played out a bit with IoT and voice controlled systems like Alexa. They've got their places, but nobody needs or wants the Amazon Dash buttons, or for Alexa to do your shopping for you.

Setting an alarm or adding a note to a list is fine, remote monitoring is fine, but when it comes to things that really matter like spending money autonomously, it completely falls flat.

Long story short, I see a fad that will fall into the background of what people actually do, rather than becoming the medium that they do it by.

[throwout4110]

I could not disagree more but you are far from alone and I respect a lot of the reasons I’ve gathered why you and others have this belief

[skywhopper]

Maybe those people do different work than you do? Coding agents don’t work well in every scenario.

[throwout4110]

Yet people imply that because it doesn’t work in their scenario that it’s not good?