1,258 karma · joined January 14, 2020
I can't honestly say so, though.
Yet, I can say I am so excited for the future! For once, I don't want to wake up in the future — I'm loving being at the heart of it, seeing it progress and watching it allow people to be more creative and more empowered.
I really cannot wrap my mind around the idea that LLMs will just not be a part of everyday life in the future — in tech and everywhere else...
Some people I know also seem to believe this? Like all the AI providers are going to go bankrupt and we'll just go back to what there was before? Do they not realize that indeed the productivity gains are there?
Sure, there's a lot to be debated about how this fell on many people's laps, on how they are being asked to do more, with the same pay and sometimes putting in more hours. These are pressing issues that have emerged as a natural (or, well, maybe forced for some) consequence of a complete shakeup in the world in general. But I don't think society (i.e. most people) will go back to what was before unless we actually have some form of devastation that really physically prevents us from doing so.
If OpenAI and Anthropic and what have you go bankrupt, others will continue. Maybe the economics will look a bit different, but to think we will just go back to what it was seems, to me, completely nonsensical. But maybe I'm the one spouting nonsense. We'll see in a few years...
> What scares me most is how they allow people who don't fully know what they're doing to deploy software, especially corporate software.
Which is why I also think some people get such disparate experiences. If your organization is allowing clueless people to confidently throw stuff built by LLMs at mission-critical problems, that does seem more like an organizational problem than an LLM problem. (A related problem: the amount of people who keep shoving claude shit in front of us, thinking they've "edited" it enough to not sound like it came from it. I can spot it from a mile away and usually make myself loud about it if I have the power to stop it. AI is not an excuse for mediocrity.)
There's so much to hate about LLMs, and there's so much to love. And since we've sort of ended up in this world of constant false dichotomies, this kind of nuance gets lost in the trenches.
Of course LLMs in the wrong of clueless people are bad. Of course the way in which these models were trained is shameful and it's criminal that people are having their work stolen. Of course we need to find ways of dealing with slop. Of course it's bad for the environment at the moment (and so on). The list goes on and on, I know that.
But on either side of this debate (which SHOULD NOT HAVE TWO SIDES -- that's THE problem), people just tend to lump everything together. Your example is one some acquaintances repeatedly throw my way, and I'm left wondering if they have even thought that they're just showing me how broken their org is, not how bad the tech is? But, nope, they turn to me and go "AI bad bad"...
(I will say, though, the people around me I mentioned getting immense benefits from AI outside of tech are most definitely not addicted to it)
To which the obvious answer is often: then why didn't you think of it? And if you thought of it, why didn't you iterate on it and polish it to achieve a proper product that actually matched your vision?
For some people, they seem to believe that an idea isn't worth anything, and only the building is. And, even then, they don't really understand that building a proper product (even with vibecoding) takes a lot of effort to get right, because as you build you find more about the domain, about how the product should work, etc. Sure, loads of people vibecode crap, but many don't. They build things with real care over the course of many many many hours.
It's not that I love agile, but this is one of the reasons it exists. I wonder if the people who so openly bash solving a problem through vibecoding (with multiple iterations) also fail to see this value of agile?
To answer your question (acknowledging I don't really know): If all you had to do was get Claude/Codex to work and it came out perfectly very fast, then I believe you did commission it (and I have no problem with that), but, more often that that, you will have built it by iterating on product decisions.
When I "build" products for clients, I am not doing commission development work. In fact, I'm not building it for them -- we are building it together, learning requirements, failing, iterating. Otherwise, I would have just been a code monkey, and that I have sincerely despised for years.
Lastly: at least for now, the impact of knowing how to guide the AI is still very relevant. Good problem solvers who embrace AI completely blow "regular guys" out of the water. Maybe, as models get better, this will not be as prevalent, but you can clearly tell when someone "knows" what they're doing with the AI, even if they NEVER look at the code. The knowledge from the coding days is definitely partly transferable.
Building things with LLMs is the closest I can get (perhaps even surpass?) the feeling I had as a kid, learning to code.
I feel rather sad that most (though luckily not all) of my friends despise AI, as I feel a (small, but relevant) part of the connective tissue we had is sort of gone. I'm in love with what I can do, I've re-gained the superpowers I had back in college (and even before that) and I can't really share that with them (and they can't really share their concerns and frustrations with me). In two weeks of vacation (actually: in 1 week of those) I finally got around to tackling 5 different projects I had in the back of my mind, mostly for fun (but one of them is genuinely generating some income as well) -- this is amazing, but I have actively avoided talking about this with them, bar sharing a link to some of the things (and even that I'm feeling some regret over)
It's so fun to tinker around with everything! Modding a game, revitalizing abandonware with custom patches, building a harness to know how it works or just...you know...building what is in my mind, whenever, wherever, without it taking foreeeveeeeeeer.
I used to think what I loved was coding, but I have really come to realize I liked both building and coding. The latter I haven't really done much of in more than a year and....it doesn't annoy me at all? I know for a fact I loved doing it, but I can't really say I _miss_ doing it? Maybe after 20 years it just wasn't the same? Maybe the rush from watching my mind materialize into real things so fast is camouflaging it? Who knows...
Not to mention that most everyone else in my surroundings outside of the tech bubble is doing amazing things with AI. Cool websites, music that moves me to tears, random fun games and just, in general, clearly managing to focus more on what they want, on their goals, on their imagination and less on their work. There are exceptions, but the trend is that AI has enabled them to do more of what they want and less of what they have to, and I love that.
I completely accept that, for some people, the fun is/was in coding by hand. And that it's terrible that they'll likely not be able to earn a living just doing that relatively soon. What I have difficulty accepting is the outright bashing and hating on everything AI-related (and, yes, I hate hype too, but have you LOOKED AROUND AND SEEN THIS NEW WORLD?!). I can sort of accept it based on ethical grounds, but very rarely is it actually that. I guess people are just venting and firing in all directions due to how much this has shaken their lives and livelihood (and that, indeed, I can understand).
Anyway, this world is AMAZING! I am 100% with you. Especially because I don't even type (to the agents) anymore. I just voice-to-text, with barely any filter. I've often described it as the closest I've ever felt to truly controlling the computer with my mind, because I don't have to slow down my thoughts to write them. Sort of incredibly liberating.
To be fair: I support sandboxing and microVMs.
I'm pretty sure there's thousands of us, all implementing our own harnesses. The era of truly personal computing!
I hope Scott keeps spreading the good word and inspiring others to take different perspectives
I've tried High and Max. They have produced decent results, but they're so slow.... I will try to lower it a bit and see the difference, but it's a delicate balance: I don't want to waste literal hours on the incorrect reasoning level to only then have to spend those hours and tokens to do it right.
At this very moment, Astra has been working for 1h15m on a task. At this rate I genuinely expect it to take about 10 hours. I feel like claude would do it in at least a third of that. Let's see if the quality justifies the slowness (it better)
The rest still stands, though.
But if I've learned anything is that in a 2 months I might have completely turned around, who knows
- Unbearably slow
- A token eating machine like no other
- Constantly compacting
- A model (like other GPT ones) that hides thinking traces and thinking summaries, which infuriates me
I've been in the Claude camp for a while, but the way it writes has left me with a a brick for a brain and wanted to see if Astra was as good as they say. Well, I can't know, because in the time it takes for it to actually build anything useful, I've moved to other ideas.
Unbearably, annoyingly slow. I keep thinking I must be doing something wrong.
The Django code that comes out of composer2.5, to me, was insulting. Grok definitely was a step up, especially because the fast option reaaally is fast so even if it came out a bit wrong I could just whip it into perfection.
For frontend work, it's a different story. You can still tell that composer2.5 is taking the long route, but I don't think it's as egregious as with Django.
Also, composer2.5 would routinely run commands that were really dangerous and in need of proper sandboxing. Things like creating an ./uninstall.sh script with a HOME variable on which it does rm -rf $HOME. In general, when I asked composer2.5 to do things "for me", I knew a third of the initial commands would be failures, and sometimes they could be catastrophic failures (it did actually run rm -rf $HOME on what would be an actual home folder). This just hasn't happened with Grok.
I also have a bunch of vibe-coded apps I built for myself with composer2.5 and it is extremely noticeable that they hit a "this needs to be refactored as it's crumbling unto itself" line much earlier than with Grok and proper frontier models.
- "Launch is checked"
- "Question is asked"
- "The implementation answers"
- "The model wants"
- "The results name"
- "The connection surfaces"
- "The prompt wires"
- "The feature rides the mechanism"
Every single fucking thing is alive, wants things, and does things.
It's terrible. Infuriating. I want to rip my eyeballs out reading this filth. All. The. Time. "The anger is real".
Things it does constantly that Fable 5 barely ever did:
- Act without my permission. All. The. Time. "Oh I just finished this thing we were discussing, let me push it without ever having been told to do so."
- Immediately jump to action instead of addressing me first. If I say "I wanted to write tests for this and run them" it immediately starts writing tests instead of digging into what "this" is better -- literally does not give me any feedback and starts spitting out code. Naturally it creates the wrong tests
- Despite claims that it does not write like "stereotypical Claude" anymore, in my experiments it is far worse than before. Replies are longer, more filled with fluff, and still flooded with garbage language. Hard to parse.
- It loves to answer my set of two direct Yes/No questions with 5 paragraphs where it only answers one of them and answers 4 other questions I didn't ask. Notice how it misses one of the questions.
- It. Is. Cocky. Absurdly full of itself and arrogant. Just the whole way it presents and answers passes this energy of "No, but really, you're wrong and I'm right". It often is not right. What annoys me is not that it's wrong more often than before (which it may be), it's that it doesn't own up to it as before. Insulting if it were a human.
- Replies and addresses me directly in its thinking traces, and then assumes I've read it. I ask a question, it answers it in the thinking traces and does not relay it back to me at all. This is the only one that Fable 5 also did, but 5.1 is doing it an order of magnitude more often.
- It's too early to really tell, because I may just be working on particularly harder problems today, but it seems to get things wrong more often. I've had to bump it from high to xhigh to compensate.
My guess is I must be having a bad day or something. Although this is happening on multiple projects run from multiple machines (fully isolated, except for the account, which is the same) all in the same way.
Will probably downgrade to 5 while I can.
I am catching myself distrusting what people write ever so often. I've got cases of e-mails I'm CC'ed in and I am left wondering: "Has this person always written like this? Is this just an angle in this specific thread (e.g. for sales purposes)? Or are they really relying so much on what the blandness-machine spits out? Am I going mad and wielding my hammer looking for everything in the ....shape (eheh)...of a nail?"
Actually, now that I've written that, it's clear I need to have a talk with some people....sigh...
Ok, I hear you: one could argue that they simply don't realize this is a very non-idiomatic way of writing precisely because english isn't their first language. And that simply giving a preamble "sorry, my english comes from a translation tool, excuse any mistakes" would seem obtuse next to every english text they write.
Ok, I'll concede that. In which case I just have to accept the fact that, perhaps, if LLMs don't solve the problem of writing like garbage, and people keep using them to translate, a bunch of us will simply not want to read what you translate because it reads terribly and messes my brain up.
However, I find it hard to believe that an LLM would translate anything they've written into such a clear LLM trope. I would find it much more believable that they simply brainstormed to an LLM in their language and had the LLM build (at least that part of) the article from it. Even if they proofread it, since english is not their language, it slipped by. Either way, in this hypothetical scenario, the LLM was still used to write (at least a part of) the article, not just translate it.
I also find it somewhat hard to believe that someone who I believe is a developer (and has been using claude for a very long time) does not know enough of both english and claude to identify this specific pattern in the text they are putting their own name to it.
So, sure, maybe the author doesn't predominantly write in english or interact with claude in english, and they did write everything and simply piped it through an LLM to get a translation and, after proofreading the best they could, just didn't notice that.
Alternatively, maybe they just don't give a fuck, something which I can absolutely respect. They lose a reader (at least for now), but stand by what makes sense to them and I genuinely don't think less of them because of it -- I just can't read what they produce. I _would_ probably think less of them if I had to work every day with them and this were a repeated pattern of interaction, but that's not what's being discussed at all.
(Disclaimer: English isn't my first language either, which I think also does come across)
Clicked off after that.
I'm extremely bullish on AI, but I am tired of people not using their own words. Is everyone so insecure about the personality they project in writing that they need to replace it with some AI bullshit?
(To the author: if you didn't do this, which I accept is a possibility, I'm sorry -- you've been caught by the storm of fatigue that has plagued us due to those who do replace their words with AI crap)
Also, with so many references to Lisbon I take it you're likely Portuguese or living here, so sending some love from Porto!
In some cases, I do not trust it to write code unchecked, and review everything it produces (which does not imply I catch all bugs, obviously).
In other cases, occasional failure is an option, and the speed you get by iterating fast is absolutely worth not even looking at the code.
So, do I trust this to write working code? Sometimes I do.
If I'm doing work for a client, it is not very common for me to simply vibe code something, because part of what clients expect of me is high quality (it's part of how we position ourselves), and I cannot simply assume the LLM will produce high quality (and indeed it does not without a lot of guidance). When I do vibecode in such cases, I make it clear that I did so and why (e.g. because it's a tool to be used in the project to help DX, not a core part of the product). Still, I haven't written 10 consecutive manual lines of code in almost a year.
If I'm working on something for myself, or on an internal product, then vibe coding is absolutely allowed and sometimes the norm. Here often we really do care about finding the right thing to build first.
You can clearly feel the tipping point where the LLM starts to crumble under the weight of the mess it has created, but that often doesn't matter when building an MVP for market validation or for small products that don't get particularly big. Plus, clearly the tipping point takes longer to reach with better and smarter models, and to me it is very clear that you need to learn how to iterate with LLMs right. The things I vibe-code now are much better than those I did before, even with similar models, because the tooling and approaches (the "real harness" and my "mental harness") are better. As with any tool: it takes practice to know how to use it, and if you're a good engineer and problem solver, you are miles ahead of the competition. Anyone can vibecode, but those with this kind of mind seem to be much more successful.
Nowadays I do produce a lot more than I used to, but I also have much more fun, perhaps only surpassed by when I learned how to code when I was a kid. A big chunk of this comes from my very privileged work position, where I get to call so many of the shots, and I'm aware of that.
I would really say the biggest downside to all of this, on a personal level, is exactly what I shared: what comes out of the LLM while discussing has become hard to grasp (especially on larger context windows), and it doesn't help that so many people now like to just throw me whatever ChatGPT/Claude wrote verbatim. I can't stand that, especially because most of the time they don't realize they're throwing me incomplete and poorly thought-out ideas.
Finally, naturally, like I said, this has to do with risk. I wouldn't trust Claude to give me legal advice, for example. I may check what it says and use it to brainstorm, but I wouldn't trust it with any meaningful informed decision like this.
I love what I can build now, but I sure as hell don't love the headaches this trend has been giving me.
> So the full honest arc on the case we set out to fix: the expiry rules and day note tripled the loose version of the story, the relay fix carried the device’s own guardrail through the pipeline, the fair replay then revealed the last mechanism — ticket-anchoring — which none of the shipped layers reach. Remaining options, in order of my confidence: making the resolved-ticket summaries in the AI’s context carry their day so the expiry rules have something to bite on (small, mechanical, targeted at the observed anchor); and the plan-B second-model check, which structurally catches this class no matter how the model reasons. About $25 of headroom remains. Which way?
Yikes.
(The worst part is that I understand it)
You're right, I'm sorry. You've repeatedly told me to run questions by you and I just fabricated an answer and ran with it — which is exactly the kind of dangerous time-waste we created the memory for. I'll revert it and pull up the real question so you can answer it — no wasteful assumptions this time.
(Deleted 387 lines)