Every 2/3 months we're hearing there's a new model that just blows the last one out of the water for coding. Meanwhile, here I am with Opus and Sonnet for $20/mo and it's regularly failing at basic tasks, antigravity getting stuck in loops and burning credits. We're talking "copy basic examples and don't hallucinate APIs" here, not deep complicated system design topics.

It can one shot a web frontend, just like v0 could in 2023. But that's still about all I've seen it work on.

If you hired a human, it will cost you thousands a week. Humans will also fail at basic tasks, get stuck in useless loops, and you still have to pay them for all that time.

For that matter, even if I'm not hiring anyone, I will still get stuck on projects and burn through the finite number of hours I have on this planet trying to figure stuff out and being wrong for a lot of it.

It's not perfect yet, but these coding models, in my mind, have gotten pretty good if you're specific about the requirements, and even if it misfires fairly often, they can still be useful, even if they're not perfect.

I've made this analogy before, but to me they're like really eager-to-please interns; not necessarily perfect, and there's even a fairly high risk you'll have to redo a lot of their work, but they can still be useful.

> salesman can convince executives that it does

I tend to think that reality will temper this trend as the results develop. Replacing 10 engineers with one engineer using Cursor will result in a vast velocity hit. Replacing 5 engineers with 5 "agents" assigned to autonomously implement features will result in a mess eventually. (With current technology -- I have no idea what even 2027 AI will do). At that point those unemployed engineers will find their phones ringing off the hook to come and clean up the mess.

Not that unlike what happens in many situations where they fire teams and offshore the whole thing to a team of average developers 180 degrees of longitude away who don't have any domain knowledge of the business or connections to the stakeholders. The pendulum swings back in the other direction.

Maybe I shouldn't have used the word "replaced", as I don't really think it's actually going to "replace" people long term. I think it's likely to just lead to higher output as these get better and better .

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

[2] https://en.wikipedia.org/wiki/Gustafson%27s_law

Anecdotally I’ve seen no difference in model changes in the last year, but going from LLM to Claude code (where we told the LLMs they can use tools on our machines) was a game changer. The improvement there was the agent loop and the support for tools.

In 2023 I asked v0.dev to one shot me a website for a business I was working on and it did it in about 3 minutes. I feel like we’re still stuck there with the models.

Of course, sample size of one, so if you haven't gotten those results then fair enough, but I've at least observed it getting a lot better.

Now it's reversed. More often than not its method is better than mine (e.g. leveraging a better function/library than I would have).

In general, it's writing idiomatic mode much more often. It's been many months since I had to correct it and tell it to be idiomatic.

There were definitely languages where it worked better (JS), but if I told people here I had to spend a lot of time tweaking after it, at least half of them assumed I was being really anal about spacing or variable names, which was simply not the case.

It’s still the case for cheaper models (GPT-mini remains a waste of my timetime), but there’s mid level models like Minimax M2 that can produce working code and stuff like Sonnet can produce usable code.

I’m not sure the delta is enough for me that I’d pay for these tools on my own though…

We all know LLMs get stuck. We know they hallucinate. We know they get things wrong. We know they get stuck in loops.

There are two types of people: The first group learns to work within these limits and adapt to using them where they’re helpful while writing the code when they’re not.

The second group gets frustrated every time it doesn’t one-shot their prompt and declares it all a big farce. Meanwhile the rest of us are out here having fun with these tools, however limited they are.

> The whole discourse around LLMs is so utterly exhausting. If I say I don't like them for almost any reason, I'm a luddite. If I complain about their shortcomings, I'm just using it wrong. If I try and use it the "right" way and it still gets extremely basic things wrong, then my expectations are too high.

As I’ve said, I use LLMs, and I use tools that are assisted by LLMs. They help. But they don’t work anywhere near as reliably as people talk about them working. And that hasn’t changed in the 18 months since I first promoted v0 to make me a website.

Using LLMs has made it fun for me to make software again.

Overconfident and over-positive shallow posts just hurt the overall discussion. Also some layer of arrogance - a typical 'if you struggle to get any significant value out of this new toy you must be doing something horribly wrong, look at us all being 100x productive!' which is never ever followed by some detailed explanation of their stack and other details.

Clearly the tools have serious issues since most users struggle to get any sustained reliable added value, and everybody keeps hoping things will improve later due to it being able to write lengthy prose on various topics or fill our government documents.

Yeah, kids these days just include stdio.h and start printing stuff, no understanding of register allocation or hardware addressing modes. 20 years from now nobody will know how to write an operating system.

Also some layer of arrogance

As compared to "if you claim AI is useful for you, you're either delusional or a shill"? The difference is that the pro-AI side can accept that any specific case it may not work well, while detractors have to make the increasingly untenable argument that it's never useful.

If your metric is an LLM that can copy/paste without alterations, and never hallucinate APIs, then yeah, you'll always be disappointed with them.

The rest of us learn how to be productive with them despite these problems.

I struggle to take comments like this seriously - yes, it is very reasonable to expect these magical tools to copy and paste something without alterations. How on earth is that an unreasonable ask?

The whole discourse around LLMs is so utterly exhausting. If I say I don't like them for almost any reason, I'm a luddite. If I complain about their shortcomings, I'm just using it wrong. If I try and use it the "right" way and it still gets extremely basic things wrong, then my expectations are too high.

What, precisely, are they good for?

The gist is this: Programmers view computers as deterministic. They can't tolerate a tool that behaves differently from run to run. They have a very binary view of the world: If it can't satisfy this "basic" requirement, it's crap.

Programmers have made their career (and possibly life) being experts at solving problems that greatly benefit from determinism. A problem that doesn't - well either that needs to be solved by sophisticated machine learning, or by a human. They're trained on essentially ignoring those problems - it's not their expertise.

And so they get really thrown off when people use computers in a nondeterministic way to solve a deterministic problem.

For everyone else, the world, and its solutions, are mostly non-deterministic. When they solve a problem, or when they pay people to solve a problem, the guarantees are much lower. They don't expect perfection every time.

When a normal human asks a programmer to make a change, they understand that communication is lossy, and even if it isn't, programmers make mistakes.

Using a tool like an LLM is like any other tool. Or like asking any other human to do something.

For programmers, it's a cardinal sin if the tool is unpredictable. So they dismiss it. For everyone else, it's just another tool. They embrace it.

[1] This, of course, is changing as they become better at coding.

I use LLMs, I code with a mix of antigravity and Claude code depending on the task, but I feel like I’m living in a different reality when the code I get out of these tools _regularly just doesn’t work, at all_. And to the parents point, I’m doing something wrong for noticing that?

> And to the parents point, I’m doing something wrong for noticing that?

There's nothing wrong pointing out your experience. What the OP was implying was he expects them to be able to copy/paste reliably almost 100% of the time, and not hallucinate. I was merely pointing out that he'll never get that with LLMs, and that their inability to do so isn't a barrier to getting productive use out of them.

> he'll never get that with LLMs, and that their inability to do so isn't a barrier to getting productive use out of them.

This is _exactly_ what the comment thread we're in said - and I agree with him. > The whole discourse around LLMs is so utterly exhausting. If I say I don't like them for almost any reason, I'm a luddite. If I complain about their shortcomings, I'm just using it wrong. If I try and use it the "right" way and it still gets extremely basic things wrong, then my expectations are too high.

> If it were terrible, you wouldn't use them, right? Isn't the fact that you continue to use AI coding tools a sign that you find them a net positive? Or is it being imposed on you?

You're putting words in my mouth here - I'm not saying that they're terrible, I'm saying they're way, way, way overhyped, their abilities are overblown, (look at this post and the replies of people saying they're writing 90% of code with claude and using AI tools to review it), but when we challenge that, we're wrong.

Ah, no. This is wildly off the mark, but I think a lot of people don't understand what SWEs actually do.

We don't get paid to write code. We get paid to solve problems. We're knowledge workers like lawyers or doctors or other engineers, meaning we're the ones making the judgement calls and making the technical decisions.

In my current job, I tell my boss what I'm going to be working on, not the other way around. That's not always true, but it's mostly true for most SWEs.

The flip side of that is I'm also held responsible. If I write ass code and deploy it to prod, it's my ass that's gonna get paged for it. If I take prod down and cause a major incident, the blame comes to me. It's not hard to come up with scenarios where your bad choices end up costing the company enormous sums of money. Millions of dollars for large companies. Fines.

So no, it has nothing to do with non-determinism lol. We deal with that all the time. (Machine learning is decades old, after all.)

It's evaluating things, weighing the benefits against the risks and failure modes, and making a judgement call that it's ass.

Then there are the cases where I just cannot get it do what I ask. Ask Gemini to remove the background of an image and you get a JPEG with a backed in checkerboard background, even when you tell it to produce an RGBA PNG. Again, I don't have any use for that.

But it does know a lot of things, and sometimes it informs me of solutions I was not aware of. The code isn't great, but if I were non-technical (or not very good), this would be fantastic and better than I could do.

I agree that there's definitely some overhype to them right now. At least for the stuff I've done they have gotten considerably better though, to a point where the code it generates is often usable, if sub-optimal.

For example, about three years ago, I was trying to get ChatGPT to write me a C program to do a fairly basic ZeroMQ program. It generated something that looked correct, but it would crash pretty much immediately, because it kept trying to use a pointer after free.

I tried the same thing again with Codex about a week ago, and it worked out of the box, and I was even able to get it to do more stuff.

scamming people

It is a helper, a partner, it is still not ready go the last mile

If you hired a staff engineer to sit next to me, and I just had him/her write 100% of the code and never tried to understand it, that would be an unwise decision on my part and I'd have little room to complain about the times he made mistakes.

> The whole discourse around LLMs is so utterly exhausting. If I say I don't like them for almost any reason, I'm a luddite. If I complain about their shortcomings, I'm just using it wrong. If I try and use it the "right" way and it still gets extremely basic things wrong, then my expectations are too high.

I’m perfectly happy to write code, to use these tools. I do use them, and sometimes they work (well). Other times they have catastrophic failures. But apparently it’s my failure for not understanding the tool or expecting too much of the tool, while others are screaming from the rooftops about how this new model changes everything (which happens every 3 months at this point)

That experience gave me a deep appreciation for how incredible LLMs are and the amazing software they can power—but it also completely demystified them. So by all means, let’s use them. But let’s also understand there are no miracles here. Go back to Shannon’s papers from the ’60s, and you'll understand that what seems to you like "emerging behaviors" are quite explainable from an information theory background. Learn how these models are built. Keep up with the latests research papers. If you do, you’ll recognize their limitations before those limitations catch you by surprise.

There is no silver bullet. And if you think you’ve found one, you’re in for a world of pain. Worse still, you’ll never realize the full potential of these tools, because you won’t understand their constraints, their limits, or their pitfalls.

See my previous comment (quoted below).

> If I complain about their shortcomings, I'm just using it wrong. If I try and use it the "right" way and it still gets extremely basic things wrong, then my expectations are too high.

Regarding "there are no miracles here"

Here are a few comments from this thread alone,

- https://news.ycombinator.com/item?id=46609559 - https://news.ycombinator.com/item?id=46610260 - https://news.ycombinator.com/item?id=46609800 - https://news.ycombinator.com/item?id=46611708

Here's a few from some older threads: - https://news.ycombinator.com/item?id=46519851 - https://news.ycombinator.com/item?id=46485304

There is a very vocal group who are telling us that there _is_ a silver bullet.

I haven't heard that at all. I hear about models that come out and are a bit better. And other people saying they suck.

>Meanwhile, here I am with Opus and Sonnet for $20/mo and it's regularly failing at basic tasks, antigravity getting stuck in loops and burning credits.

Is it bringing you any value? I find it speeds things up a LOT.

Gemini now often produces output that looks significantly better than what I could produce manually, and I'm an expert for web, although my expertise is more in tooling and package management.

I'm suspecting that once these models hit 'good enough' for ~90% of users and use cases, the providers started optimizing for cost instead of quality, but still benchmark and advertise for quality.

I think the big problem is that the pace of improvement was UNBELIEVABLE for about 4 years, and it appears to have plateaued to almost nothing.

ChatGPT has barely improved in, what, 6 months or so.

They are driving costs down incredibly, which is not nothing.

But, here's the thing, they're not cutting costs because they have to. Google has deep enough pockets.

They're cutting costs because - at least with the current known paradigm - the cost is not worth it to make material improvements.

So unless there's a paradigm shift, we're not seeing MASSIVE improvements in output like we did in the previous years.

You could see costs go down to 1/100th over 3 years, seriously.

But they need to make money, so it's possible non of that will be passed on.

I have no idea what this is called, but it feels like a lot of people assume that progress will continue at a linear pace for forever for things, when I think that generally progress is closer to a "staircase" shape. A new invention or discovery will lead to a lot of really cool new inventions and discoveries in a very short period of time, eventually people will exhaust the low-to-middle-hanging fruit, and progress kind of levels out.

I suspect it will be the same way with AI; I don't now if we've reached the top of our current plateau, but if not I think we're getting fairly close.

I'll interpret it as meaning 1800s to 1900s to 2000s. I'd argue that we haven't yet seen the same step change as 1800s to 1900s this century because we're only just beginning the ramp up on the new technology that will drive progress this century similar to how in 1926 they were still ramping up on the use of electricity and internal combustion engines.

Let's take electricity as the primary example though since it's the one you mentioned and it's probably more similar to our current situation with AI. The similarities include the need for central generating stations to supply raw power to end users as well as the need for products designed to make use of that power and provide some utility to the consumer. Efficiency of generation is also a primary concern for both as it's a major cost driver. Both of those required significant investment and effort to solve in the early days of electrification.

We're now solving similar problems with AI, instead of power plants we're building datacenters, instead of lightbulbs and washing machines we're developing chat bot integrations and agents, instead of improving dynamos we're improving GPUs and TPUs. I fully expect we'll follow a similar curve for deployment as we find new uses, improve existing ones and integrate this new power source into an increasing number of domains.

We do have one major advantage though, we've already built The Grid for distribution which saves a massive amount of effort.

This article is a good read on the permeation of electricity through the economy

https://www.construction-physics.com/p/the-birth-of-the-grid

We may be in a similar step-jump period now, where over the next 10-15 years we'll see some pretty big advancements in robotics due to AI, and then all of the low hanging fruit will be picked until there some other MAJOR breakthrough

Alphabet / Google doesn’t have that issue. OAI and other money losing firms do.

I'm not trying to be pedantic, but how did you arrive at 'keep improving' as a conclusion? Nobody is really sure how this stuff actually works. That's why AI safety was such a big deal a few years ago.

I will acknowledge that I don't have any evidence of this claim, so maybe the word "likely" was unwise, as that suggests probability. Feel free to replace "is "likely to" with "it feels like it will".

I made a joke once after the first time I watched one of those Apple announcement shows in 2018, where I said "it's kind of sad, because there won't be any problems for us to solve because the iPhone XS Max is going to solve all of them".

The US economy is pretty much a big vibes-based Ponzi scheme now, so I don't think we can single-out AI, I think we have to blame the fact that the CEOs running these things face no negative consequences for lying or embellishing and they do get rewarded for it because it will often bump the stock price.

Is Tesla really worth more than every other car company combined in any kind of objective sense? I don't think so, I think people really like it when Elon lies to them about stuff that will come out "next year", and they feel no need to punish him economically.

I'd rather characterize it as extremes of Greater Fool Theory.

https://en.wikipedia.org/wiki/Greater_fool_theory

The technology is neat, the people selling it are ghouls.

I know it's good tech for some stuff, just not for everything. It's the same with previous bubbles. VR is really great for some things but we were never going to work with a headset on 8 hours a day. Bitcoin is pretty cool but we were never going to do our shopping list on Blockchain. I'm just so sick of hypes.

But I do think it's good tech, just like I enjoy VR daily I do have my local LLM servers (I'm pretty anti cloud so I avoid it unless I really need the power)

It's not really about the societal impacts for me, at least not yet, it's just not good enough for that yet. I do worry about that longer-term but not with the current generation of AI. At my work we've done extensive benchmarking (especially among enthusiastic early adopters) and while it can save a couple hours a week we're nowhere near the point where it can displace FTEs.

Probably less than that, practically speaking. ChatGPT's initial release date was November 2022. It's closer to 3 years, in terms of any significant amount of people using them.

It's a business, but it won't be the thing the first movers thought it was.