2025: the Year in Llms
Key topics
As the AI landscape continues to evolve, a post reflecting on "2025: The Year in LLMs" sparked a lively discussion, with commenters waxing nostalgic about the rapid progress in the field. Some reminisced about the slower pace of development in the past, joking about the days when a year's worth of progress in Java was just a vote on adding syntactic sugar, while others fondly recalled the optimism surrounding Rust. The conversation took a humorous turn as commenters playfully debated the relative dangers of "AI bros" versus "crypto bros," with some pointing out the environmental and economic impacts of cryptocurrency mining operations in rural America. Amidst the lighthearted banter, a few commenters also touched on more serious topics, like the economic implications of large-scale AI infrastructure development.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
56m
Peak period
107
0-12h
Avg / period
22.9
Based on 160 loaded comments
Key moments
- 01Story posted
Dec 31, 2025 at 6:54 PM EST
6d ago
Step 01 - 02First comment
Dec 31, 2025 at 7:50 PM EST
56m after posting
Step 02 - 03Peak activity
107 comments in 0-12h
Hottest window of the conversation
Step 03 - 04Latest activity
Jan 5, 2026 at 2:16 AM EST
2d ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
> At the end of every month I send out a much shorter newsletter to anyone who sponsors me for $10 or more on GitHub
https://simonwillison.net/about/#monthly
I remember when we just wanted to rewrite everything in Rust.
Those were the simpler times, when crypto bros seemed like the worst venture capitalism could conjure.
I wonder what's going to cost you being this sober about AI? Your career ar minimum, I'm guessing.
And if so, what happens to those builders once the data center is built?
Yes. At some point the demand will be so high that imported workers won't suffice and local population will need to be trained and hired.
> And if so, what happens to those builders once the data center is built?
They are going to be moved to a new place where the datacenters will need to be built next. Mobility if the workforce was often cited as one of the greatest strengths of US economy.
More and more data centers (and power sources) are going to be built at the same time so more and more workers will be needed. This is going to be THE job. I think there are going to be many similarities with the age when railroads were being developed. Hopefully with less worker deaths this time.
I haven’t heard about new businesses, job creation and growth in former industrial towns. What have I missed?
This year honestly feels quite stagnant. LLMs are literally technology that can only reproduce the past. They're cool, but they were way cooler 4 years ago. We've taken big ideas like "agents" and "reinforcement learning" and basically stripped them of all meaning in order to claim progress.
I mean, do you remember Geoffrey Hinton's RBM talk at Google in 2010? [0] That was absolutely insane for anyone keeping up with that field. By the mid-twenty teens RBMs were already outdated. I remember when everyone was implementing flavors of RNNs and LSTMs. Karpathy's character 2015 RNN project was insane [1].
This comment makes me wonder if part of the hype around LLMs is just that a lot of software people simply weren't paying attention to the absolutely mind-blowing progress we've seen in this field for the last 20 years. But even ignoring ML, the world's of web development and mobile application development have gone through incredible progress over the last decade and a half. I remember a time when JavaScript books would have a section warning that you should never use JS for anything critical to the application. Then there's the work in theorem provers over the last decade... If you remember when syntactic sugar was progress, either you remember way further back than I do, or you weren't paying attention to what was happening in the larger computing world.
0. https://www.youtube.com/watch?v=VdIURAu1-aU
1. https://karpathy.github.io/2015/05/21/rnn-effectiveness/
Funny, I've used them to create my own personalized text editor, perfectly tailored to what I actually want. I'm pretty sure that didn't exist before.
It's wild to me how many people who talk about LLM apparently haven't learned how to use them for even very basic tasks like this! No wonder you think they're not that powerful, if you don't even know basic stuff like this. You really owe it to yourself to try them out.
I've worked at multiple AI startups in lead AI Engineering roles, both working on deploying user facing LLM products and working on the research end of LLMs. I've done collaborative projects and demos with a pretty wide range of big names in this space (but don't want to doxx myself too aggressively), have had my LLM work cited in HN multiple times, have LLM based github projects with hundreds of stars, appeared on a few podcasts talking about AI etc.
This gets to the point I was making. I'm starting to realize that part of the disconnect between my opinions on the state of the field and others is that many people haven't really been paying much attention.
I can see if recent LLMs are your first intro to the state of the field, it must feel incredible.
So it is absurdly incorrect to say "they can only reproduce the past."
This is even more clear in the case of diffusion models (which I personally love using, and have spent a lot of time researching). All of the "new" images created by even the most advanced diffusion models are fundamentally remixing past information. This is really obvious to anyone who has played around with these extensively because they really can't produce truly novel concepts. New concepts can be added by things like fine-tuning or use of LoRAs, but fundamentally you're still just remixing the past.
LLMs are always doing some form of interpolation between different points in the past. Yes they can create a "new" SQL query, but it's just remixing from the SQL queries that have existed prior. This still makes them very useful because a lot of engineering work, including writing a custom text editor, involve remixing existing engineering work. If you could have stack-overflowed your way to an answer in the past, an LLM will be much superior. In fact, the phrase "CRUD" largely exists to point out that most webapps are fundamentally the same.
A great example of this limitation in practice is the work that Terry Tao is doing with LLMs. One of the largest challenges in automated theorem proving is translating human proofs into the language of a theorem prover (often Lean these days). The challenge is that there is not very much Lean code currently available to LLMs (especially with the necessary context of the accompanying NL proof), so they struggle to correctly translate. Most of the research in this area is around improving LLM's representation of the mapping from human proofs to Lean proofs (btw, I personally feel like LLMs do have a reasonably good chance of providing major improvements in the space of formal theorem proving, in conjunction with languages like Lean, because the translation process is the biggest blocker to progress).
When you say:
> So it is absurdly incorrect to say "they can only reproduce the past."
It's pretty clear you don't have a solid background in generative models, because this is fundamentally what they do: model an existing probability distribution and draw samples from that. LLMs are doing this for a massive amount of human text, which is why they do produce some impressive and useful results, but this is also a fundamental limitation.
But a world where we used LLMs for the majority of work, would be a world with no fundamental breakthroughs. If you've read The Three Body Problem, it's very much like living in the world where scientific progress is impeded by sophons. In that world there is still some progress (especially with abundant energy), but it remains fundamentally and deeply limited.
You don’t have a solid background.
This is fundamentally what humans and what every function in existence does as well. It is impossible to have any sort of deterministic function, process or anything produce new information from old information. You can combine information you can transform information you can lose information. But producing new information from deterministic intelligence is fundamentally impossible.
New information can literally only arise through stochastic processes. It must be randomly generated and then selected and filtered. That’s essentially what creativity is. There is literally no other logical way to generate “new information”. Purely random is never really useful so useful information arrives only after it is filtered.
LLMs do have stochastic aspects to them so we know for a fact it is generating new things and not just drawing on the past.
The fundamental limitation with LLMs is not that it can’t create new things. It’s that the context window is too small to create new things beyond that. Whatever it can create it is limited to the possibilities within that window.
If you want to go around huffing and puffing your chest about a subject area, you kinda do fella. Credibility.
This is the internet bro. Credibility is irrelevant because identities can never be verified. So the only thing that matters is the strength and rationality of an argument.
That’s the point of hacker news substantive content not some battle of comparison of credentials or useless quips (like yours) with zero substance. Say something worth reading if you have anything to say at all, otherwise nobody cares.
Do you only take LLM seriously if it can be another Einstein?
Can you give examples of some recent, truly new, not in the past ideas?
Put another way, and I hate to throw in the now over-used phrase, but I feel you may be responding to a strawman that doesn't much appear in the article or the discussion here: "Because these tools don't achieve a god-like level of novel perfection that no one is really promising here, I dismiss all this sorta crap."
Especially when you I think you are also admitting that the technology is quite a useful tool - a stance which I think represents the bulk of the feelings that supporters of the tech here on HN are describing.
After post-training, this is definitively NOT what an LLM does.
That is a derived output. That isn't new as in: novel. It may be unique but it is derived from training data. LLMs legitimately cannot think and thus they cannot create in that way.
For a more practical example, creating bindings from dynamic-language-A for a library in compiled-language-B is a genuinely useful task, allowing you to create things that didn't exist before. Those things are likely to unlock great happiness and/or productivity, even if they are derived from training data.
This is the definition of a derived product. Call it a derivative work if we're being pedantic and, regardless, is not any level of proof that LLMs "think".
Also the derived argument doesn’t really hold, just because you know about two things doesn’t mean you’d be able to come up with the third, it’s actually very hard most of the time and requires you to not do next token prediction.
I don't believe they can. LLMs have no concept of truth.
What's likely is that the "truth" for many subjects is represented way more than fiction and when there is objective truth it's consistently represented in similar way. On the other hand there are many variations of "fiction" for the same subject.
Why is that kind of thinking required to create novel works?
Randomness can create novelty.
Mistakes can be novel.
There are many ways to create novelty.
Also I think you might not know how LLMs are trained to code. Pre-training gives them some idea of the syntax etc but that only gets you to fancy autocomplete.
Modern LLMs are heavily trained using reinforcement data which is custom task the labs pay people to do (or by distilling another LLM which has had the process performed on it).
What's clear here is that you have zero idea what you're talking about while simultaneously telling me and poorly mansplaining.
You’re using ‘derived’ to imply ‘therefore equivalent.’ That’s a category error. A cookbook is derived from food culture. Does an LLM taste food? Can it think about how good that cookie tastes?
A flight simulator is derived from aerodynamics - yet it doesn’t fly.
Likewise, text that resembles reasoning isn’t the same thing as a system that has beliefs, intentions, or understanding. Humans do. LLMs don't.
I imagine that sounded more profound when you wrote it than it did just now, when I read it. Can you be a little more specific, with regard to what features you would expect to differ between LLM and human responses to such a question?
Right now, LLM system prompts are strongly geared towards not claiming that they are humans or simulations of humans. If your point is that a hypothetical "thinking" LLM would claim to be a human, that could certainly be arranged with an appropriate system prompt. You wouldn't know whether you were talking to an LLM or a human -- just as you don't now -- but nothing would be proved either way. That's ultimately why the Turing test is a poor metric.
The mental gymnastics here is entertainment at best. Of course the thinking LLM would give feedback on how it's actually just a pattern model over text - well, we shouldn't believe that! The LLM was trained to lie about it's true capabilities in your own admission?
How about these...
What observable capability would you expect from "true cognitive thought" that a next-token predictor couldn’t fake?
Where are the system’s goals coming from—does it originate them, or only reflect the user/prompt?
How does it know when it’s wrong without an external verifier? If the training data says X and the answer is Y - how will it ever know it was wrong and reach the correct conclusion?
And beyond people claiming that LLMs are basically sentient you have people like CamperBob2 who made this wild claim:
"""There's no such thing as people without language, except for infants and those who are so mentally incapacitated that the answer is self-evidently "No, they cannot."
Language is the substrate of reason. It doesn't need to be spoken or written, but it's a necessary and (as it turns out) sufficient component of thought."""
Let that sink. They literally think that there's no such thing as people without language. Talk about a wild and ignorant take on life in general!
5 years ago a typical argument against AGI was that computers would never be able to think because "real thinking" involved mastery of language which was something clearly beyond what computers would ever be able to do. The implication was that there was some magic sauce that human brains had that couldn't be replicated in silicon (by us). That 'facility with language' argument has clearly fallen apart over the last 3 years and been replaced with what appears to be a different magic sauce comprised of the phrases 'not really thinking' and the whole 'just repeating what it's heard/parrot' argument.
I don't think LLM's think or will reach AGI through scaling and I'm skeptical we're particularly close to AGI in any form. But I feel like it's a matter of incremental steps. There isn't some magic chasm that needs to be crossed. When we get there I think we will look back and see that 'legitimately thinking' wasn't anything magic. We'll look at AGI and instead of saying "isn't it amazing computers can do this" we'll say "wow, was that all there is to thinking like a human".
Mastery of words is thinking? In that line of argument then computers have been able to think for decades.
Humans don't think only in words. Our context, memory and thoughts are processed and occur in ways we don't understand, still.
There's a lot of great information out there describing this [0][1]. Continuing to believe these tools are thinking, however, is dangerous. I'd gather it has something to do with logic: you can't see the process and it's non-deterministic so it feels like thinking. ELIZA tricked people. LLMs are no different.
[0] https://archive.is/FM4y8 [0] https://www.theverge.com/ai-artificial-intelligence/827820/l... [1] https://www.raspberrypi.org/blog/secondary-school-maths-show...
That's the crazy thing. Yes, in fact, it turns out that language encodes and embodies reasoning, if you pile up enough of it in a high-dimensional space and add some feedback in the form of RL.
No one had the faintest clue. So many people not only don't understand what just happened, they don't think anything happened at all.
What's funny is the failure to grasp any contextual framing of ELIZA. When it came out people were impressed by it's reasoning, it's responses. And in your line of defense it could think because it had mastery of words!
But fast forward the current timeline 30 years. You will have been of the same camp that argued on behalf of ELIZA when the rest of the world was asking, confusingly: how did people think ChatGPT could think?
Funnily enough, they did, if you go back far enough. It's only the deconstructionists and the solipsists who had the audacity to think otherwise.
This is the definition of the word ‘novel’.
Also , a shitton of what we do economically is reproducing the past with slight tweaks and improvements. We all do very repetitive things and these tools cut the time / personnel needed by a significant factor.
Like, I'm sorry, but you're just flat-out wrong and I've got the proof sitting on my hard drive. I use this supposedly impossible program daily.
If a programmer creating their own software (or contracting it out to a developer) would be a bespoke suit and using software someone or some company created without your input is an off the rack suit, I'd liken these sorts of programs as semi-bespoke, or made to measure.
"LLMs are literally technology that can only reproduce the past" feels like an odd statement. I think the point they're going for is that it's not thinking and so it's not going to produce new ideas like a human would? But literally no technology does that. That is all derived from some human beings being particularly clever.
LLMs are tools. They can enable a human to create new things because they are interfacing with a human to facilitate it. It's merging the functional knowledge and vision of a person and translating it into something else.
From what you've described an LLM has not invented anything. LLMs that can reason have a bit more slight of hand but they're not coming up with new ideas outside of the bounds of what a lot of words have encompassed in both fiction and non.
Good for you that you've got a fun token of code that's what you've always wanted, I guess. But this type of fantasy take on LLMs seems to be more and more prevalent as of late. A lot of people defending LLMs as if they're owed something because they've built something or maybe people are getting more and more attached to them from the conversational angle. I'm not sure, but I've run across more people in 2025 that are way too far in the deep end of personifying their relationships with LLMs.
Back to the land of reality... Describing something in fiction doesn’t magically make it "not an invention". Fiction can anticipate an idea, but invention is about producing a working, testable implementation and usually involves novel technical methods. "Star Trek did it" is at most prior art for the concept, not a blueprint for the mechanism. If you can't understand that differential then maybe go ask an LLM.
I for one think your work is pretty cool - even though I haven't seen it, using something you built everyday is a claim not many can make!
The change hit us so fast a huge number of people don’t understand how capable it is yet.
Also it certainly doesn’t help that it still hallucinates. One mistake and it’s enough to set someone against LLMs. You really need to push through that hallucinations are just the weak part of the process to see the value.
Either that, or they tried it "a year or two ago" and have no concept of how far things have gone in the meantime.
It's like they wandered into a machine shop, cut off a finger or two, and concluded that their grandpa's hammer and hacksaw were all anyone ever needed.
SWEs are trained to discard surface-level observations and be adversarial. You can't just look at the happy path, how does the system behave for edge cases? Where does it break down and how? What are the failure modes?
The actual analogy to a machine shop would be to look at whether the machines were adequate for their use case, the building had enough reliable power to run and if there were any safety issues.
It's easy to Clever Hans yourself and get snowed by what looks like sophisticated effort or flat out bullshit. I had to gently tell a junior engineer that just because the marketing claims something will work a certain way, that doesn't mean it will.
If you want a financial tip, don't short stock and chase market butterflies. Instead, make real professional friends, develop real skills and learn to be friendly and useful.
I made my money in tech already, partially by being lucky and in the right place at the right time, and partially because I made my own luck by having friends who passed the opportunity along.
Hope that helps!
The question wasn’t “are humans sometimes self-delusional?” Everyone agrees with that. The question was whether, in this specific case, the prevailing view about LLM capability is meaningfully wrong in a way that has implications. If you really believed this was mostly Clever Hans, there would be concrete consequences. Entire categories of investment, hiring, and product strategy would be mispriced.
Instead you retreated to “don’t short stocks” and generic career advice. That’s not skepticism, it’s risk-free agnosticism. You get to sound wise without committing to any falsifiable position.
Also, “I made my money already” doesn’t strengthen the argument. It sidesteps it. Being right once, or being lucky in a good cycle, doesn’t confer epistemic authority about a new technology. If anything, the whole point of contrarian insight is that it forces uncomfortable bets or at least uncomfortable predictions.
Engineers don’t evaluate systems by vibes or by motivational aphorisms. They ask: if this hypothesis is true, what would we expect to see? What would fail? What would be overhyped? What would not scale? You haven’t named any of that. You’ve just asserted that people fool themselves and stopped there.
The key point you’re missing is the type of failure. Search systems fail by not retrieving. Parrots fail by repeating. LLMs fail by producing internally coherent but factually wrong world models. That failure mode only exists if the system is actually modeling and reasoning, imperfectly. You don’t get that behavior from lookup or regurgitation.
This shows up concretely in how errors scale. Ambiguity and multi-step inference increase hallucinations. Scaffolding, tools, and verification loops reduce them. Step-by-step reasoning helps. Grounding helps. None of that makes sense for a glorified Google search.
Hallucinations are a real weakness, but they’re not evidence of absence of capability. They’re evidence of an incomplete reasoning system operating without sufficient constraints. Engineers don’t dismiss CNC machines because they crash bits. They map the envelope and design around it. That’s what’s happening here.
Being skeptical of reliability in specific use cases is reasonable. Concluding from those failure modes that this is just Clever Hans is not adversarial engineering. It’s stopping one layer too early.
Absolutely not true. I cannot express how strongly this is not true, haha. The tech is neat, and plenty of real computer scientists work on it. That doesn't mean it's not wildly misunderstood by others.
> Concluding from those failure modes that this is just Clever Hans is not adversarial engineering.
I feel like you're maybe misunderstanding what I mean when I refer to Clever Hans. The Clever Hans story is not about the horse. It's about the people.
A lot of people -- including his owner-- were legitimately convinced that a horse could do math, because look, literally anyone can ask the horse questions and it answers them correctly. What more proof do you need? It's obvious he can do math.
Except of course it's not true lol. Horses are smart critters, but they absolutely cannot do arithmetic no matter how much you train them.
The relevant lesson here is it's very easy to convince yourself you saw something you 100% did not see. (It's why magic shows are fun.)
How can anyone choose to remain so willfully ignorant in the face of irrefutable evidence that they're wrong?
https://arxiv.org/abs/2507.15855
Clever Hans was exposed because the effect disappeared under controlled conditions. Blind the observers, remove human cues, and the behavior vanished. The entire lesson of Clever Hans is not “people can fool themselves,” it’s “remove the hidden channel and see if the effect survives.” That test is exactly what has been done here, repeatedly.
LLM capability does not disappear when you remove human feedback. It does not disappear under automatic evaluation. It does not disappear across domains, prompts, or tasks the model was never trained or rewarded on. In fact, many of the strongest demonstrations people point to are ones where no human is in the loop at all: program synthesis benchmarks, math solvers, code execution tasks, multi-step planning with tool APIs, compiler error fixing, protocol following. These are not magic tricks performed for an audience. They are mechanically checkable outcomes.
Your framing quietly swaps “some people misunderstand the tech” for “therefore the tech itself is misunderstood in kind.” That’s a rhetorical move, not an argument. Yes, lots of people are confused. That has no bearing on whether the system internally models structure or just parrots. The horse didn’t suddenly keep solving arithmetic when the cues were removed. These systems do.
The “it’s about the people” point also cuts the wrong way. In Clever Hans, experts were convinced until adversarial controls were applied. With LLMs, the more adversarial the evaluation gets, the clearer the internal structure becomes. The failure modes sharpen. You start seeing confidence calibration errors, missing constraints, reasoning depth limits, and brittleness under distribution shift. Those are not illusions created by observers. They’re properties of the system under stress.
You’re also glossing over a key asymmetry. Hans never generalized. He didn’t get better at new tasks with minor scaffolding. He didn’t improve when the problem was decomposed. He didn’t degrade gracefully as difficulty increased. LLMs do all of these things, and in ways that correlate with architectural changes and training regimes. That’s not how self-deception looks. That’s how systems with internal representations behave.
I’ll be blunt but polite here: invoking Clever Hans at this stage is not adversarial rigor, it’s a reflex. It’s what you reach for when something feels too capable to be comfortable but you don’t have a concrete failure mechanism to point at. Engineers don’t stop at “people can be fooled.” They ask “what happens when I remove the channel that could be doing the fooling?” That experiment has already been run.
If your claim is “LLMs are unreliable for certain classes of problems,” that’s true and boring. If your claim is “this is all an illusion caused by human pattern-matching,” then you need to explain why the illusion survives automated checks, blind evaluation, distribution shift, and tool-mediated execution. Until then, the Hans analogy isn’t skeptical. It’s nostalgic.
Its annoying to see posts from people who lag behind in intelligence and just dont get it - people learn at different rates. Some see way further ahead.
Curious, does it perform at the limit of the hardware? Was it programmed in a tools language (like C++, Rust, C, etc.) or in a web tech?
Without you, there was nothing.
Unless you time travelled from 1946 you should be aware that there has been at least one text editor written by human that LLMs were trained on.
Is this such a big limitation? Most jobs are basically people trained on past knowledge applying it today. No need to generate new knowledge.
That's incorrect on many levels. They are drawing upon, and reproducing, language patterns from "the past", but they are combining those patterns in ways that may have never have been seen before. They may not be truly creative, but they are still capable of generating novel outputs.
> They're cool, but they were way cooler 4 years ago.
Maybe this year has been more about incremental progress with LLMs than the shock/coolness factor of talking to an LLM for the first time, but the utility of them, especially for programming, has dramatically increased this year, really in the last 6 months.
The improvement in "AI" image and video generation has also been impressive, to the point now that fake videos on YouTube can often only be identified as such by common sense rather that the fact that they don't look real.
Incremental improvement can often be more impressive that innovation, whose future importance can be hard to judge when it first appears. How many people read "Attention is all you need" in 2017 and thought "Wow! This is going to change the world!". Not even the authors of the paper thought that.
The stricter typing of Rust would make sematic errors in generated code come out more quickly than in e.g. Python because using static typing the chances are that some of the semantic errors are also type violations.
...and the best of them all, OpenCode[1] :)
[1]: https://opencode.ai
I don't see a similar option for ChatGPT Pro. Here's a closed issue: https://github.com/sst/opencode/issues/704
I like to believe, but MCP is quickly turning into an enterprise thing so I think it will stick around for good.
MCP is a great way for an LLM to connect to an external system in a standardized way and immediately understand what tools it has available, when and how to use them, what their inputs and outputs are,etc.
For example, we built a custom MCP server for our CRM. Now our voice and chat agents that run on elevenlabs infrastructure can connect to our system with one endpoint, understand what actions it can take, and what information it needs to collect from the user to perform those actions.
I guess this could maybe be done with webhooks or an API spec with a well crafted prompt? Or if eleven labs provided an executable environment with tool calling? But at some point you're just reinventing a lot of the functionality you get for free from MCP, and all major LLMs seem to know how to use MCP already.
I don't think MCP is going to go away, but I do think it's unlikely to ever achieve the level of excitement it had in early 2025 again.
If you're not building inside a code execution environment it's a very good option for plugging tools into LLMs, especially across different systems that support the same standard.
But code execution environments are so much more powerful and flexible!
I expect that once we come up with a robust, inexpensive way to run a little Bash environment - I'm still hoping WebAssembly gets us there - there will be much less reason to use MCP even outside of coding agent setups.
Planing depends on deterministic view of the future. I used to plan (esp annual plans) until about 5 years. Now I scan for trends and prepare myself for different scenarios that can come in the future. Even if you get it approximately right, you stand apart.
For tech trends, I read Simon, Benedict Evans, Mary Meeker etc. Simon is in a better position make these predictions than anyone else having closely analyzed these trends over the last few years.
Here I wrote about my approach: https://www.jjude.com/shape-the-future/
Here is the changelog for OpenBSD 7.8:
https://www.openbsd.org/78.html
There's nothing here that says: We make it easier to use it more of it. It's about using it better and fixing underlying problems.
Mistakes and hallucinations matter a whole lot less if a reasoning LLM can try the code, see that it doesn't work and fix the problem.
Does it? It's all prompt manipulation. Shell script are powerful yes, but not really huge improvement over having a shell (REPL interface) to the system. And even then a lot of programs just use syscalls or wrapper libraries.
> can try the code, see that it doesn't work and fix the problem.
Can you really say that does happens reliably?
If you mean 100% correct all of the time then no.
If you mean correct often enough that you can expect it to be a productive assistant that helps solve all sorts of problems faster than you could solve them without it, and which makes mistakes infrequently enough that you waste less time fixing them than you would doing everything by yourself then yes, it's plenty reliable enough now.
Its very difficult to argue the point that claude code:
1) was a paradigm shift in terms of functionality, despite, to be fair, at best, incremental improvements in the underlying models.
2) The results are an order of magnitude, I estimate, better in terms of output.
I think its very fair to distill “AI progress 2025” to: you can get better results (up to a point; better than raw output anyway) without better models with clever tools and loops. (…and video/image slop infests everuthing :p).
My point is purely that, compared to 2024, the quality of the code produced by LLM inference agent systems is better.
To say that 2025 was a nothing burger is objectively incorrect.
Will it scale? Is it good enough to use professionally? Is this like self driving cars where the best they ever get is stuck with an odd shaped traffic cone? Is it actually more productive?
Who knows?
Im just saying… LLM coding in 2024 sucked. 2025 was a big year.
It’s also possible that people more experienced, knowledgable and skilled than you can see fundamental flaws in using LLMs for software engineering that you cannot. I am not including myself in that category.
I’m personally honestly undecided. I’ve been coding for over 30 years and know something like 25 languages. I’ve taught programming to postgrad level, and built prototype AI systems that foreshadowed LLMs, I’ve written everything from embedded systems to enterprise, web, mainframes, real time, physics simulation and research software. I would consider myself an 7/10 or 8/10 coder.
A lot of folks I know are better coders. To put my experience into context: one guy in my year at uni wrote one of the world’s most famous crypto systems; another wrote large portions of some of the most successful games of the last few decades. So I’ve grown up surrounded by geniuses, basically, and whilst I’ve been lectured by true greats I’m humble enough to recognise I don’t bleed code like they do. I’m just a dabbler. But it irks me that a lot of folks using AI profess it’s the future but don’t really know anything about coding compared to these folks. Not to be a Luddite - they are the first people to adopt new languages and techniques, but they also are super sceptical about anything that smells remotely like bullshit.
One of the most wise insights in coding is the aphorism“beware the enthusiasm of the recently converted.” And I see that so much with AI. I’ve seen it with compilers, with IDEs, paradigms, and languages.
I’ve been experimenting a lot with AI, and I’ve found it fantastic for comprehending poor code written by others. I’ve also found it great for bouncing ideas. And the code it writes, beyond boiler plate, is hot garbage. It doesn’t properly reason, it can’t design architecture, it can’t write code that is comprehensible to other programmers, and treating it as a “black box to be manipulated by AI” just leads to dead ends that can’t be escaped, terrible decisions that will take huge amounts of expert coding time to undo, subtle bugs that AI can’t fix and are super hard to spot, and often you can’t understand their code enough to fix them, and security nightmares.
Testing is insufficient for good code. Humans write code in a way that is designed for general correctness. AI does not, at least not yet.
I do think these problems can be solved. I think we probably need automated reasoning systems, or else vastly improved LLMs that border on automated reasoning much like humans do. Could be a year. Could be a decade. But right now these tools don’t work well. Great for vibe coding, prototyping, analysis, review, bouncing ideas.
What are some of the models you've been working with?
they were right
Invariably they've never used AI, or at most very rarely. (If they used AI beyond that, this would be admission that it was useful at some level).
Therefore it's reasonable to assume that you are in that boat. Now that might not be true in your case, who knows, but it's definitely true on average.
- fart out demos that you don't plan on maintaining, or want to use as a starting place
- generate first-draft unit tests/documentation
- generate boilerplate without too much functionality
- refactor in a very well covered codebase
It's very useful for all of the above! But it doesn't even replace a junior dev at my company in its current state. It's too agreeable, makes subtle mistakes that it can't permanently correct (GEMINI.md isn't a magic bullet, telling it to not do something does not guarantee that it won't do it again), and you as the developer submitting LLM-generated code for review need to review it closely before even putting it up (unless you feel like offloading this to your team) to the point that it's not that much faster than having written it yourself.
https://news.ycombinator.com/newsguidelines.html
Different strokes, but I’m getting so much more done and mostly enjoying it. Can’t wait to see what 2026 holds!
Anyone that believes that they are completely useless is just as deluded as anyone that believes they're going to bring an AGI utopia next week.
2024 was a lot of talk, a lot of "AI could hypothetically do this and that". 2025 was the year where it genuinely started to enter people's workflows. Not everything we've been told would happen has happened (I still make my own presentations and write my own emails) but coding agents certainly have!
431 more comments available on Hacker News