AI coding is gambling
notes.visaint.space
notes.visaint.space
On the surface this does not sound as satisfying, because it more resembles shopping than coding. But once Claude Code is finally tuned to do its job perfectly, you will essentially be using that infinite app store. You're actually using it right now, every time you use Claude Code — just an imperfect version of it.
If you enjoy using AI because it allows you to "will anything into existence", it's because the process is currently imperfect. Using Claude Code is closer to shopping than coding, but because the process is obfuscated, it feels like you're the one making the products in the shopping catalogue every time you place an order.
the LLM equivalent would be to prompt "give me an app", without specifying what that app does and then repeating that until you get the app you are looking for, each time, checking by hand if the app does what you want.
They may be innate, but that doesn't mean they are related or that one is a good substitute for the other.
When I walk down a street, I get 10 people stopping me to ask "Where did you get that?". When I tell them I made it, their heads explode. I know which side of that interaction is more satisfying.
We also go all-out for Halloween, and at the big Halloween festival there is literally a line down the street of people waiting to take photos with us. We created something amazing.
People aren't going to line up for slop.
You really believe that? What has lead you to the conclusion that LLMs will ever be capable of that?
For example, if a carpenter was given a perfect hammer, or a painter a perfect paintbrush, would they find their craft any less enjoyable? AI, on the other hand, falls into a different category of tools (if we can call them tools at all) since they would no longer be enjoyable as tools of creation once they reached their "perfect" state.
Typing is just choosing from the latent space something special, too. Could just be random words, or, even fewer, random grammatically correct sentences.
It's just one step along the path of AI adoption to execute on an idea and see in near real-time the idea you had baked in your head come alive in front of you. Most of us get to this point and become the biggest evangelists of the tech. I see no reason you should feel guilty for the excitement you're feeling right now, and you should enjoy the journey. You're definitely paying for it in tokens, that's for sure.
However, there will come a point at which you will have successfully willed into existence a novel thing that you always wanted, and there it is, exactly as you dreamed, but by then, you'll be left with a weird empty feeling you won't really have the words for. Maybe it's a feeling of not earning the thing you built, or maybe it's just, your idea is finished and now you have to think of another idea. Certainly, this was your idea though, and it proves you were right, or at least on to something, and it is valuable, to you.
Yet, you didn't go on the journey to get there. You didn't bump up against limitations of the programming language or system and think about workarounds while you were showering or commuting to the office. You basically bought the finished product from the dynamic template marketplace of Anthropic (or whereever), and that's cool that it does what you need. It just isn't really programming, or being a software engineer in the traditional sense.
What used to be something you could potentially leave your day job for to go create a startup with a cofounder over, or maybe sell off to a buyer, or just open source and share with the world, isn't going to have the same meaning. It's a black box of code that you'll need a coding agent to continue working on, keeping that money flowing to Anthropic or whereever.
Anyway, I think the Slot Machine question is where a lot of early adopters are now at in this journey, and once more of us are there, then we can start asking the hard questions. Right now too many of us are where you're at, and it's impossible to know where things will end up in a year or so.
I want computers to do work for me. Any barrier between me and the result I want is an annoyance.
It's very weird because judging from this comment, and some other comment you wrote asking whether the other person believed creation requires hard work (which wasn't at all what they said), makes it seem as if you aren't reading the comments you're replying to.
You do not like and have never liked programming. You wanted to be a manager. They are completely different things.
it seems you and others took my words a bit more literally than I intended for them to come across. it's not like I'm just one-shotting all my ideas directly into existence, I still need to understand how to use the tool to do it. it's just a different tool. one that's allowing me to build way more than I ever have, while having a ton of fun doing it.
and sure, your analogy seems reasonable if I was simply buying the code w/ my tokens. that wouldn't be fun or fulfilling at all - it's more like there is some new "cooking" tool that immediately spawns 90% of the ingredients pre-cut & prepped (maybe the other 10% isn't exactly what I asked for but I can improvise with it) and gives me a decent recipe based on the idea of what I wanted to cook in the first place that fills in (and gives me a starting point to learn about) the gaps that I didn't even realize I was missing. I see it more as: "All this time I thought I loved chopping onions and setting up the grill, but actually I just loved cooking".
you weren't wrong about the mcdonalds though. I do love mcdonalds
They love the craft, for all they care they could be working in a black box in a void as long as it fed them interesting problems to solve.
They don't see any actual benifit in the AI increasing the velocity of how fast they build useful things. That was never of value to them, all they see is the problems becoming more boring to solve.
I believe AI will do something similar for programming. The level of complexity in modern apps is high and requires the use of many technologies that most of us cannot remotely claim to be expert in. Getting an idea and getting a prototype will definitely be easier. Production Code is another beast. Dealing with legacy systems etc will still require experts at least for the near future IMHO.
Developers will always disagree on the best tool for X ... but we should all fear the Luddites who refuse to even try new tools, like AI. That personality type doesn't at all mesh with my idea of a "good programmer".
All the excellent developers around me are _not_ using AI except for very small, contained tasks.
I am sure that you're objectively wrong if that is what you're saying.
in my world they are called product managers or product owners (scrum) but they are not programmers. prompting an LLM is producing a product but it is not programming.
i refuse to use AI because i want to remain a programmer, and not become a manager.
Ai is tool, that lets us do what we want to do.
I also remember trying to create my first iOS app in Xcode and thinking «this is simply beyond me». Wouldn’t say app coding is trivial with ai, but its at least feasible now.
The bad part that none of us have a competitive edge anymore, and are close to unemployable. We can’t all be self taught founders who starts our own businesses. It’s going to get weird.
You could just as easily make claims about carpentry or cooking because you discovered Ikea or microwave meals. They serve a purpose and technically satisfy the needs of anyone, yet they aren't a good enough solution for anything important. That's where we're at with this tech.
Nobody is selling this off a shelf. The market for "Eloquence-compatible formant synthesizer with a citation-backed parameter space" is approximately me.
I could not have built this without AI coding tools, and I've been a professional developer for 15+ years. I wrote the specs, I chose the architecture, I read the papers, I know what the output needs to sound like. The sheer surface area of translating hundreds of papers worth of acoustic phonetics into a working runtime would have taken me years solo. With Claude Code it's taken months, and I'm still the one catching when it misinterprets a Klatt coefficient or botches a formant transition rule, because you have to actually know the domain to do that.
Reducing what you do not understand to "Ikea or microwave meals?" Because you don't like it? or aren't familiar with it? Is saying a thing about you. Not about people who know how to use the tools.
I'm saying there's a ton of nuance and human feedback necessary to build the software most developers work on for a living. It's built to requirements that evolve with the business. Businesses ultimately serve people with opinions and preferences. Businesses need to pass audits. Specs can change quarterly. Clients and their contracts come and go. It's exactly like building/maintaining custom furniture for a bespoke house, or consistently cooking a signature recipe at any scale and considering any necessary accommodations. If it wasn't true, the business wouldn't be viable regardless if AI is used or not. I'm talking about systems and services, not products.
You are building a product for a narrow use case requiring DSP. Modeling was always the point. I don't doubt AI helped, but we're not talking about the same thing.
If you were building something for an enterprise client, nobody would give a shit how you got it done as long as you have a demo by monday and it ships next month. No excuses and no gotchas. Any incident risks breaking the SLA and losing the contract. If the client calls a meeting at the last minute to make changes, you're probably working on the weekend. People don't like using AI for stuff like that. Efficiency is not the priority. People don't want cutting edge or novel. They want reliability and competence. Their livelihood depends on knowing exactly how it works and how it can be extended and maintained. They want to stay at least one step ahead of what the client asks for next. People want to test the hell out of it and nobody wants to get fired for not noticing what is obvious to other stakeholders who don't share their tunnel vision. Clients don't have much tolerance for delays or bugs. This is why AI has mixed or negative results for all but personal projects or startups.
This is exactly the sort of mentality that makes me hate this technology
You finally feel good at programming despite admitting that you aren't actually doing it
Please explain why anyone should take this seriously?
I agree with gp that the speed in which I am able to execute my vision is exhilarating. It is making me love programming again. My side projects, which have been hanging on the wall for years, are actually getting done. And quickly!
The actual act of keying in code is drudgery for me. I've written so much code in so many languages that it is hard not to hate them all. Why the fuck is it a hash in ruby but a dict in python? How the hell do I get the current unixtime in this language again?!? Why the fuck do I need to learn yet another stupid vocabulary for what is essentially databinding? Who cares, let the AI handle it
These are the downsides, but there are also upsides like in human languages: “wow I can express this complex idea with just these three words? I never though about that!”. Try a new programming paradigm and that opens your mind and changes your way of programming in _any_ language forever.
I believe gp and others just like food instead of cooking. Which is fine, but if that's the case, why go around telling everyone you're a cook?
If I get a robot someday and manage it daily before I leave for work to slowly build a house, when it's done, I gotta be honest and admit I'll consider myself a home builder.
Otherwise, who is a home builder? Very few people do every single part themselves, even if they technically could.
It didn't mean we shouldn't use C++ and stop hand-writing (almost all) assembly. I don't think it means we should't use LLMs and stop hand-writing C++ either.
OP defines it as getting the machine to do as he wants.
You define it as the actual act of writing the detailed instructions.
Programming is willing the machine to do something... Writing code is just that writing code, yes sometimes you write code to make the machine do something and other times you write code just to write code ( for example refactoring, or splitting logic from presentation etc.)
Think about it like this... Everyone can write words. But writing words does not make you a book writer.
What always gets me is that the act of writing code by itself has no real value. Programming is what solves problems and brings value. Everyone can write code, not everyone can "program"....
I agree with OP because the journey itself rarely helps you focus on system architecture, deliverable products and how your downstream consumers use your product. And not just product in the commercial sense, but FOSS stuff or shareware I slap together because I want to share a solution to a problem with other people.
The gambling fallacy is tiresome as someone who, at least I believe, can question the bullshit models try to do sometimes. It is very much gambling for CEOs, idea men who do not have a technical floor to question model outputs.
If LLMs were /slow/ at getting a working product together combined with my human judgement, I wouldn't use them.
So, when I encounter someone who doesn't pin value into building something that performs useful work, only the actual journey of it, regardless of usefulness of said work, I take them as seriously as an old man playing with hobby trains. Not to disparage hobby trains, because model trains are awesome, but they are hubris.
Hyperbole, yes, many things are in fact, not possible. But most people have the size of the two categories confused. The number of things that are categorically impossible is less than a rounding error compared to how many things are possible.
The joy and wonder of being an engineer is in taking problems deemed "impossible" and creating possibilities. It's in extracting a solution from infinite possibilities and redefining what possible even is.
> The number of things that are categorically impossible is less than a rounding error compared to how many things are possible.
If it's just a case of keeping a positive attitude and self-help, I can accept it. A sort of white lie one tells themselves.
I kept at it on the side as a hobby. But stacks evolved and I was left behind. Now with AI it's back on.
i learned programming in high school and i enjoyed it, then while starting computer science i did and internship at a software company, and i hated it, and i thought i hated programming and wanted to give up studying computer science, but when i discovered programming MUDs and then web development with the same language i loved it again. turns out i always loved programming, what i didn't like was the corporate work environment, 9-5, using CASE tools (remember those?) on windows, maybe the feeling of inferiority as an untrained intern among everyone else.
what i hate about LLMs is the tediousness, the unreliability, having to try over and over to get a result.
i often work with customers directly, less technical ones too. seeing their satisfaction when i solve a problem for them (no matter how) is what allowed me to keep going doing even non-programming work, though i admit that i prefer programming if that customer interaction is missing. so i too love the art, but i still love problem solving if there is someone who appreciates the solution.
so maybe it wasn't problem solving that was your problem, but the big business environment, and how it constrained your role and didn't give you the feedback you needed?
Yes, it is insane. You couldn't torture this confession out of me. But that's the drug they're selling you, isn't it? You don't even write code, but you're getting a self-inflated sense of worth. It must be addicting! Of course, whether or not the programs you prompt are actually good surely has no relation to whether you feel they're good, since you're not the one writing them, and apparently were not capable of writing them before so are not qualified to review them very much.
> having tools that can finally match the speed my ideas come to me
Anyone can be an "ideas guy". We laughed at those people, because having ideas is not the hard part. The hard part was in all of the hundreds and thousands of little details that go into building the ideas into something actually worthwhile, and that hasn't changed. LLMs can build an idea into a prototype in a weekend. I am still waiting to see LLMs build an idea into something other people use at scale, once, ever, other than LLM wrappers. Either every person who is all-in on vibes only has ideas that consist of making .md files and publishing them as a "meta agent framework", or LLMs are not actually doing a great job of translating ideas into tangibly useful software.
I disagree with this. I've worked with amazing "ideas guys" who just cranked out customer insights and interesting concepts, and I've worked with lousy ones, who just kinda meandered and never had a focused vision beyond a milquetoast copy of the last thing they saw. There's a real skill to forming good concepts, and it's not a skill everyone has!
I've been using AI tools more but this idea of never actually writing any code seems way too black and white to be serious.
I think there's way more nuance to this than you're willing to admit here. There's a significant difference between the guy who thinks "I'm going to make X app to do Y and get loaded." and the person who really understands the details of what they want to create and has a concrete vision of how to shape it.
I think that product shaping and detail oriented vision of how something should work and be used by people is genuinely challenging, wholly aside from the lower level technical skills required to execute it.
This is part of the reason why I wouldn't be surprised at all to see product manager types getting more hands-on, or seeing the software engineering profession evolve into more of a PM/SDE hybrid.
Here is the source code for a greenfield, zero-dependency, 100% pure PHP raw Git repository viewer made for self-hosted or shared environments that is 99.9% vibe-coded and has had ~10k hits and ~7k viewers of late, with 0 errors reported in the logs over the last 24 hours:
Sure it's easy to create bad ideas. Not easy at all to create good ones.
Just to nitpick, because I think the difference is relevant: "Idea to prototype in a weekend" was possible for a spirited coder already before LLMs.
Now it's "Idea to prototype in a few minutes".
That’s because when it comes to delivering value, code doesn’t matter: outcomes do.
If I spend 10 hours hand coding something versus prompting an LLM to create a solution that delivers the same outcome in a few minutes, and I can get that solution into production in under an hour from the moment my fingers first touch the keyboard to start writing the prompt, well, whilst these solutions might both deliver the same value, the ROI differs significantly.
I agree. It's the "buy in" from the market.
The biggest names in Software Products have (other peoples) ideas to sell, they're selling the buggy versions of those ideas - Microsoft, Salesforce, even early Facebook, these weren't triumphs of 'monk-like discipline' in the code. They were triumphs of market buy in and timing.
"I am still waiting to see LLMs build an idea into something other people use at scale" - so Microsoft using Claude Code doesn't count?
I don’t think I’d describe my behavior as destructive though
I've ended up with a process that produces very, very high quality outputs. Often needing little to no correct from me.
I think of it like an Age of Empires map. If you go into battle surrounded by undiscovered parts of the map, you're in for a rude surprise. Winning a battle means having clarity on both the battle itself and risks next to the battle.
> AI coding is gambling on slot machines, managing developers is gambling on the stock market.
Because I feel like that is a much more apt analogy.
And the quality of code models puts out is, in general, well below the average output of a professional developer.
It is however much faster, which makes the gambling loop feel better. Buying and holding a stock for a few months doesn't feel the same as playing a slot machine.
https://simonwillison.net/2025/Feb/3/a-computer-can-never-be...
E.g. look at the "SWE-Bench Pro (public)" heading in this page: https://openai.com/index/introducing-gpt-5-4/ , showing reasoning efforts from none to high.
Of course, they don't learn like humans so you can't do the trick of hiring someone less senior but with great potential and then mentor them. Instead it's more of an up front price you have to pay. The top models at the highest settings obviously form a ceiling though.
My experience is the absolute opposite. I am much more in control of quality with Ai agents.
I am never letting junior to midlevels into my team again.
In fact, I am not sure I will allow any form of manual programming in a year or so.
Except, one can explain themselves (humans) and their actions can be held to account in the case of any legal issue whereas an AI cannot; making such an entity completely unsuitable for high risk situations.
This typical AI booster comparison has got to stop.
Employees can only be held accountable with severe malice.
There is a good chance that the person actually responsible (eg. The ceo or someone delegated to be responsible) will soon prefer to have AIs do the work as their quality can be quantified.
You "own" the software it creates which means you're responsible for it. If you use AI to commit crimes you'll go to jail, not the AI.
>You are not an AI and do not know how an AI "thinks".
>Even if you come to be able to anticipate an AI's output, you will be undermined by the constant and uncontrollable update schedule imposed on you by AI platforms. Humans only make drastic changes like this under uncommon circumstances, like when they're going through large changes in their life, not as a matter of course.
>However, without this update schedule, problems that were once intractable will likely stay so forever. Humans, on the other hand, can grow without becoming completely unpredictable.
It's a Catch-22. AI is way closer to gambling.
These guys had to manage very complex calculation engine based on we’ll just let it changes every year had to be correct had to be delivered by a certain date every year.
They had an army (100-200 people depending on various factors) of marginally skilled coding drones that were able to turn out the Java, COBOL or whatever it was predictably on that schedule without necessarily understanding any of the big picture or have any having any hope of so. Basically a software factory. There was about a dozen people who actually understood everything.
It wasn't a real game of hangman, it was flat out manipulation, engagement farming. Do you think it's possible that AI does that in any other situations?
So, it technically didn't change the secret word so much as it was trying to infer what its own secret word might have been, based on your guesses.
As @m00x points out "coding is gambling on slot machines, managing developers is betting on race horses."
Being a project manager is more or less something humans have been doing since the dawn of time.
Generative AI takes money as input and gives some output. If you don’t like the output, more money goes in. It’s far more akin to gambling than organizing human labor.
Heck, this technology also offers a parasocial relationship at the same time! Plopping tokens into a slot-machine which also projects a holographic "best friend" that gives you "encouragement" would fit fine in any cyberpunk dystopia.
I believe that's the strongest pattern in LLM gambling. Was listening the Syntax and they described that "Even though theLLM did it wrong 4 times, that 5th time could be right, so why not just go!"; paraphrased of course.
It also explains the meta-LLM business, where all these CEO types put in some question and because the LLM just knows all these words, they believe it's valuable because it's "almost" correct, even when that last correction might be forever elusive because these machines arn't thinking, they're patterning a highly regularized language beneath the more loose descriptions.
There'll definitely be a winner in the AI bubble, but it'll be seen after it pops.
This probably won't surprise anyone familiar with Japanese corporate culture: external pressure to boost productivity just doesn't land the same way here. People nod, and then keep doing what they've always done.
It's a strange scene to witness, but honestly, I'm grateful for it. I've also been watching plenty of developers elsewhere get their spirits genuinely crushed by coding agents, burning out chasing the slot machine the author describes. So for now, I'm thankful I still get to see this pastoral little landscape where people just... write their own code.
But where there are no tests, and you're the one defining whats correct, you're definitely encroaching on the slot machine hoping it'll spit out something faster than you could do it yourself.
Then there's some vague unease in whether spending the time to prompt will actuall result in a properly integrated software in a large existing code base with idiosyncratic code use.
Overall, I don't see the ROI business gets from forcing people into these tools; however, as an individual, it's definitely worth understanding what they can do. Mostly, I see the efficient copy/past/find/replace of existing code to be very good.
Know when to Re-prompt,
Know when to Clear the Context,
And know when to RLHF.
You never trust the Output,
When you’re staring at the diff view,
There’ll (not) be time enough for Fixing,
When the Tokens are all spent.
Bold assumption that people are looking at the diffs at all. They leave that for their coworkers agents.
[1]: https://web.archive.org/web/20230130060050/https://www.rolli...
I'm not an expert in the psych/neuro literature on addiction, but I suspect latency isn't that critical. But is that just because it's things like fruit-machines that have been studied? Gambling (poker, racehorses) are quite long-latency. OTOH, scrolling is closer to 400ms, and that's certainly the modern addition...
It is exactly like the proverbial monkey or rat pressing a bar for a food pellet to come out.
If the pellet unerringly drops, and is always tasty and nutritious, the rat stops when it's no longer hungry.
Otherwise, an inordinate amount of time is spent pressing the bar.
Why do you often need to re-prompt things like "can you simplify this and make it more human readable without sacrificing performance?". No amount of specification addresses this on the first shot unless you already know the exact implementation details in which case you might as well write it yourself directly.
I often have to put in a prompt like this 5-10 times before the code resembles something I'd even consider using as a 1st draft base to refactor into something I would consider worthy of being git commit.
I sometimes use AI for tiny standalone functions or scripts so we're not talking about a lot of deeply nested complexity here.
Are you stuck entering your prompts in manually or do you have it setup like a feedback loop like "beautify -> check beauty -> in not beautiful enough beautify again"? I can't imagine why everyone things AIs can just one shot everything like correctness, optimization, and readability, humans can't one shot these either.
I have one shot prompted projects from empty folder to full feature web app with accounts, login, profiles, you name it, insanely stable, maybe and oops here or there, but for a non-spec single prompt shot, that's impressive.
When I don't use a tool to handle the task management I have Claude build up a markdown spec file for me and specify everything I can think of. Output is always better when you specify technology you want to use, design patterns.
This seems like 1980's corporate waterfall thinking, doesn't jibe with the messy reality I've seen with customers, unclear ideas, changing market and technical environments, the need for iteration and experimentation, mid-course correction, etc.
The PMs asked ChatGPT to write a well-formed spec.
Sadly, true in too many companies right now.
I do agree with your general point that The Spec can become a crutch for washing your hands of any responsibility for knowing the product, the goals, the company's business, and other contexts. I like to defuse these ideas by reminding the engineers that The Spec is a living document and they are partially responsible for it, too. Once everyone learns that The Spec isn't a crutch for shifting all blame to the product manager, they become more involved in making sure it's right.
That's where the gambling metaphor really resonates. It's not whether or not the output is correct, I've been building software for many years and I know how direct LLMs pretty well at this point. But I'm also an alcoholic in recovery and I know that my brain is wired differently than most. And using LLMs has tested my ability to self-regulate in ways that I haven't dealt with since I deleted social media years ago.
I dont think i've read a sentence on this website i can relate to less.
I watch the LLM build things and it feels completely numb, i may as well be watching paint dry. It means nothing to me.
But even accounting for all these "hard" constraints and metrics, there are clearly reasons to prefer some possible programs over others even when they all satisfy the same constraints and perform equally on all relevant metrics.
We do treat programs as efficient causes[1] of side effects in computing systems: a file is written, a block of memory is updated, etc. and the program is the cause of this.
But we also treat them as statements of a theory of the problem being solved[2]. And this latter treatment is often more important socially and economically. It is irrational to be indifferent to the theory of the problem the program expresses.
Maintainability is a big one missing from the current LLM/agentic workflow.
When business needs change, you need to be able to add on to the existing program.
We create feedback loops via tests to ensure programs behave according to the spec, but little to nothing in the way of code quality or maintainability.
[Imports the completely fabricated library docker_quantum_telepathy.js and calls the resolve_all_bugs_and_make_coffee() method, magically compiling the code on an unplugged Raspberry Pi]
AI: "Done! The production deployment was successful, zero errors in the logs, and the app works flawlessly on the first try!"
Just instead of hitting keys, they’re hitting words, and the words have probability links to each other.
Who the hell thinks this is ready to make important decisions?
Which with the advent of LLMs just lowered our standards so we can claim success.
The endless next steps of "and add this feature" or "this part needs to work differently" or "this seems like a bug?" or "we must speed up this part!" is where 98% of the effort always was.
Is it different with AI coding?
1/ Dependency -- Once I got used to agentic coding, I almost always reached out to it even for small changes (e.g. update a yaml config)
2/ Addiction -- In the initial euphoria phase, many people experience not wanting to "waste" any time agent idle and they'd try to assign AI agents task before they go to sleep
3/ You trust your judgement less and less as agent takes over your code
4/ "Slot machine" behavior -- running multiple AI agents parallel on same task in hope of getting some valuable insight from either
5/ Psychosis -- We have all met crypto traders who'd tell you how your 9-5 is stupid and you could be making so much trading NFTs. Social media if full of similar anecodotes these days in regards to vibecoding with people boasting their Claude spend, LOC and what not
It's not an inherent feature to slot machines, it's something we enforce because people got angry about the outcomes (i.e. fraud) when they didn't operate that way.
It doesn't matter because a dodgy slot-machine is still a slot machine, and the person using it would still be a gambler.
You can get more consistent results from a slot machine with a bunch of magnets and some swift kicks. It's still gambling.
The important part of the not-really-a-metaphor is the relationship between user and machine, and how it affects the user's mind.
What the machine outputs on "wins" doesn't matter as much, addictive gambling can still happen even when the payouts are dumb.
This is a subreddit about selfhosting things others built for free. Honestly, often for piracy purposes. It's insane how entitled people have become.
We love a good holy war for sure.
The nuance is lost, and the conversations we should be having never happen (requirements, hiring/skills, developer experience).
That isn't true, which is the exact reason why people have a binary mindset. More than once on Hacker News I've had people accuse me of being an AI booster just because I said I had success with agents and they did not.
Personally I use coding agents for boring parts (I really don't enjoy putting the same piece of string to 20 different classes just to register a new component) and they work quite well, I'm going to use them for foreseeable future, because they make coding much more enjoyable for me. On the other hand I don't have an OpenClaw box burning billions of tokens weekly for me, because I usually don't have ideas that could be clearly specified.
Applies here? :D
The risk isn't randomness per se it's over trusting something that looks correct. The skill ceiling is moving from "can you write it" to "can you reliably verify it"
Good programmers might have made things that “performed well”, and had “few bugs”, without this step, but it was not robust to changes over time. If we end up in a place where every project has solid automated verification, perhaps things get better overall.
I am not going to say what it is because all of the AI haters will immediately flock to leave it bad reviews and overwhelm my support systems with bad faith requests (something that has already happened).
I've been writing software for 25 years, I know what I am doing. Every bug I shipped was my fault either because I didn't test well enough or I did not possess enough platform knowledge to know myself the right way to do things. "Unknown unknowns"
But I have also learned better ways to do things and fixed every bug using AI tools. I don't read the code. I may scan it to gain context and then tweak a single value myself, but beyond that I don't write or read code anymore.
Its not a magical few shot prompt then reap profits machine. I just feel like a solopreneur ditch digger who just got a lease on a new CAT excavator. I can get work done faster I can also do damage faster if I am not careful.
Beyond this concern,
I’ve certainly been spending more time coding. But is it because it’s making me more efficient and smarter or is it because I’m just gambling on what I want to see?
Am I just pulling the lever until I reach jackpot?
At the very least, a gambler knows when they have hit jackpot. Here, you start off assuming you've won the jackpot every time, and maybe there'll be an unpleasant surprise down the line. Maybe that's still gambling, but it's pretty backwards.
Fast & Cheap (but not Good?) - I wouldn't really say that AI coding is "cheap"
Cheap & Good (but not Fast) - Again, not really "cheap"
Fast & Good (but not Cheap) - This seems like maybe where we're at? Is this a bad place?
Eventually, it will be just Fast and Good. It won't be cheap, as companies start moving towards profitability.
Remember when Uber was super cheap? I do. They're fast and good though.
Overall I’m a fan, but yes there are things to watch for. It doesn’t replace skilled humans but it does help skilled humans work faster if used right.
The labor replacement story is bullshit mostly, but that doesn’t mean it’s all bad.
Now, the job is to nail the spec and test HARD against that spec. Let the AI develop it and question it along the way to make sure it's not repeating itself all over the place (even this I'm sure is super necessary anymore...). Find a process that helps you feel comfortable doing this and you can get the engineering part done at lightning speed.
Both jobs are scary in different ways. I find this way more fun, however.
2. Who here thinks that having interns write all/almost all of your code and moving all your mid level and senior developers to exclusively reviewing their work and managing them is a good idea?
Coding agents look at existing text in the codebase before they act. If they previously used a pattern you dislike and you tell them how to do differently, the next time they run they'll see the new pattern and are much more likely to follow that example.
There are fancier ways of having them "learn" - self-updating CLAUDE.md files, taking notes in a notes/ folder etc - but just the code that they write (and can later read in future sessions) feels close-enough to "learning" to me that I don't think it makes sense to say they don't learn any more.
I’ve never worked anywhere where the interns had net productivity on average.
The reason i think this metaphor keeps popping up, is because of how easy it is to just hit a wall and constantly prompt "its not working please fix it" and sometimes that will actually result in a positive outcome. So you can choose to gamble very easily, and receive the gambling feedback very quickly unlike with an intern where the feedback loop is considerably delayed, and the delayed interns output might simply be them screaming that they don't understand.
The first is equating human and LLM intelligence. Note that I am not saying that humans are smarter than LLMs. But I do believe that LLMs represent an alien intelligence with a linguistic layer that obscures the differences. The thought processes are very different. At top AI firms, they have the equivalent of Asimov's Susan Calvin trying to understand how these programs think, because it does not resemble human cognition despite the similar outputs.
The second and more important is the feedback loop. What makes gambling gambling is you can smash that lever over and over again and immediately learn if you lost or got a jackpot. The slowness and imprecision of human communication creates a totally different dynamic.
To reiterate, I am not saying interns are superior to LLMs. I'm just saying they are fundamentally different.
And, if we're being honest, the way people talk about interns is weirdly dehumanizing, and the fact that they are always trotted out in these AI debates is depressing.
Yeah, I agree with that.
That thought crossed my mind as I was posting this comment, but I decided to go with it anyway because I think this is one of those cases where I think the comparison is genuinely useful.
We delegate work to humans all the time without thinking "this is gambling, these collaborators are unreliable and non-deterministic".
You should value assigning tasks to human interns more than AI because they are human
But looks like the intern mafia is bombarding you with downvotes.
Addiction and recovery is part of my story, so I've done quite a bit of work around that part of my life. I don't gamble, but I can confidently say that using LLMs has been an incredible boost in my productivity while completely destroying my good habits around setting boundaries, not working until 2AM, etc.
In that sense, it feels very much like gambling.
The idea that I would only be able to work with an internet connection and a subscription to a big LLM provider is not very appealing to me. And I'm not even talking about the ethical concerns...
I suppose what's happening with software development is we're exploring where the line between the two is going to land. It's pretty clear that something like a simple and generic website can be reliably vibecoded, but on the other extreme I wouldn't expect the software for something like a space shuttle to be vibe coded due to the stringent safety requirements.
Sometimes I think we put the Carr before the horse. We gamble because evolution promotes that approach.
Yes I could go for the reliable option. But taking a punt is worth a shot if the cost is low.
The cost of AI is low.
What is a problem is people getting wrapped up in just one more pull of the slot machine handle.
I use AI often. But fairly often I simply bin its reponse and get to work on my own. A decent amount of the time I can work with the response given to make a decent result.
Sometimes, rarely, it gives me what I need right off the bat.
If we're only talking about money spent on prompting AI, maybe. The damage to online trust is massive imo. So is the damage done by looting the commons to build them.
Typical privatize the profits socialize the costs bullshit
Important to point out that that every high culture produced restrictions on exactly those behaviors, gambling was a universal vice when that concept still mattered.
America in particular had a work culture that favored well, work and technical excellence. Now work is for suckers, thinking is for suckers, precision not worth it when you can have some machine do it half-right.
"Yes I could go for the reliable option. But taking a punt is worth a shot if the cost is low.", might as well be the national slogan from vibe-coding to the department of defense. Even the venture capital industry that excels at slot machine sectors was itself already a slot machine.
Humans invented gambling as a rigged game that mimics what's in nature, perversed for profit.
You need to collect food, do you go to where you know there are berries (low value but high likelihood of finding), or scout off to find a herd of deer? (High value but low likelihood of finding).
Looking for deer wouldnt be walking off in a random direction. You check water holes, known clearings, known fields.
Each of these is an operation (walk to X and look), each has a low probability of meeting a deer.
This is a variable reward scheme.
The result is optmize foraging practices - you mostly hunt for deer then fall back to berries. In larger groups some will gather berries some will hunt.
Contrary to popular thought hunter and gatherer were not separate occupations.
AI has removed some of the tedium, and freed up more of my bandwidth to think about the problems I’m trying to solve and what the actual best ways to solve those problems are.
Only once I have a good feel for the problem I am solving do I go to the AI for help implementing.
My style of prompting usually leads to code that is very close to what I would have manually typed. I review it and tweak it until it is effectively identical to what I would have typed.
The speed up is significant. YMMV.
For any one hand you may win or lose, but on average you should still take the "gamble" every time, because the odds are so good you'll win. That's what using AI is like right now.
This is a new state of things - a year ago, the analogy would have been reversed.
it's really extremely similar to working with a junior programmer
so in this post, where does this go wrong?
> I am not your average developer. I’ve never worked on large teams and I’ve barely started a project from scratch. The internet is filled with code and ideas, most of it freely available for you to fork and change.
Because this describes a cut-and-paster, not a software architect. Hence the LLM is a gambling machine for someone like this since they lack the wisdom to really know how to do things.
There's of course a huge issue which is that how are we going to get more senior/architect programmers in the pipeline if everyone junior is also doing everything with LLMs now. I can't answer that and this might be the asteroid that wipes out the dinosaurs....but in the meantime, if you DO know how to write from scratch and have some experience managing teams of programmers, the LLMs are super useful.
Right, which is why LLMs aren't useful if you actually know what you're doing. It's a drain on your time to have to carefully check everything a junior writes, but you do it because he will learn and eventually return on that investment. With an LLM, there is no such long term payoff.
A big theme of software development for me has been finishing things other people couldn’t finish and the key to that is “control variance and the mean will take care of itself”
Alternately the junior dev thinks he has a mean of 5 min but the variance is really 5 weeks. The senior dev has mean of 5 hours and a variance of 5 hours.
This has been how I think about it, too. The success rates are going up, but I still view the AI as an adversary that is trying to trick me into thinking it's being useful. Often the act is good enough to be actually useful, too.
Now you have more resources to test, reduce permissions scope, to build a test bench & procedure. All of the excuses you once had for not doing the job right are now gone.
You can write 10k + lines of test code in a few minutes. What is the gamble? The old world was a bigger gamble.
Lmk how you feel when you're constantly build integrations with legacy software by hand.
But you can't survive alone on an island. So you need to determine where to make gambles and where not.
Watching vibe gamblers hooked onto coding agents who can't solve fizz buzz in Rust are given promotional offers by Anthropic [0] for free token allowances that are the equivalent in the casino of free $20 bets or free spins at the casino to win until March 27, 2026.
The house (Anthropic) always wins.
[0] https://support.claude.com/en/articles/14063676-claude-march...
Defining “Gambling” like isn’t really helpful.
You can't keep paying to play the "refinancing game" until you get a good rate (at least not like pulling the lever again and again, you have to wait a long time, you won't call the same bank again and again, and suddenly they have an amazing rate), it's a different experience and the psychology is different.
If someone only ever paid $20 a month on slot machines AND typically came out ahead, that wouldn’t be any kind of gambling issue.
Playing real slot machines is only really a problem if you regularly put in too much because the odds are literally set against you. Not true with LLMs, both you and the AI company would become more successful the more helpful their tool is.
Is.
Life.
You've discovered probability, there was an 80% change of that. Roll a dice and do not pass go.
Again. The output from llm is a probable solution, not right, not wrong.
If, on the other hand, you treat it like a hyper-competent collaborator, and follow good project management and development practices, you're golden.
I am consistently using 100% of my weekly $200 max plan. I know how this thing works, I know how to get value out of it, and I wish what you said were true.
If you do all of these things? You are in a better spot. You are in a far better spot than if you hadn't! Setting up hooks to ensure notes get written? Massive win! Red-green TDD? Yes, please! But in terms of just ... well, being able to rely on the damn thing?
In a healthy environment. We are harmed more by being totally risk adverse. Than by accepting risk as part of life and work.
It also depends on what you're coding with;
- If you're coding with opus4.6, then it's not gambling for a while.
- If you'r coding with gemini3-flash, then yeah.
One thing I have noticed though is- you have to spend a lot of tokens to keep the error/hallucination rate low as your codebase increases in size. The math of this problem makes sense; as the code base has increased, there's physically more surface where something could go wrong. To avoid that you have to consistently and efficiently make the surface and all it's features visible to the model. If you have coded with a model for a week and it has produced some code, the model is not more intelligent after that week- it still has the same layers and parameters, so keeping the context relevant is a moving target as the codebase increases (and that's why it probably feels like gambling to some people).
> you have to spend a lot of tokens to keep the error/hallucination rate low
Ironically, I find your comment more effective at convincing me AI coding is gambling than the original article. You're talking about it the exact same way that gamblers do about their games.
- Was there anymore intelligence that you wanted to add to your argument?
When I have Claude create something from scratch, it all appears very competent, even impressive, and it usually will build/function successfully…on the surface. I have noticed on several occasions that Claude has effectively coded the aesthetics of what I want, but left the substance out. A feature will appear to have been implemented exactly as I asked, but when I dig into the details, it’s a lot of very brittle logic that will almost certainly become a problem in future.
This is why I refuse to release anything it makes for me. I know that it’s not good enough, that I won’t be able to properly maintain it, and that such a product would likely harm my reputation, sooner or later. What frightens me is there are a LOT of people who either don’t know enough to recognize this, or who simply don’t care and are looking for a quick buck. It’s already getting significantly more difficult to search for software projects without getting miles of slop. I don’t know how this will ultimately shake out, but if it’s this bad at the thing it’s supposedly good at, I can only imagine the kinds of military applications being leveraged right now…
That’s only half of the transition.
The other half - and when you know you’ve made it through the “AI sux” phase - is when you learn to automate the mopping up. Give the agent the info it needs to know if it did good work - and if it didn’t do good work, give it information so it knows what to fix. Trust that it wants to fix those things. Automate how that info is provided (using code!) and suddenly you are out of the loop. The amount of code needed is surprisingly small and your agent can write it! Hook a few hundred lines of script up to your harness at key moments, and you will never see dumb AI mistakes again (because it fixed them before presenting the work to you, because your script told it about the mistakes while you were off doing something else)
Think of it like linting but far more advanced - your script can walk the code AST and assess anything, or use regex - your agent will make that call when you ask for the script. If the script has an exit code of 2, stderr is shown to the agent! So you (via your script) can print to stderr what the agent did wrong - what line, what file, wha mistake.
It’s what I do every day and it works (200k LOC codebase, 99.5% AI-coded) - there’s info and ideas here: https://codeleash.dev/docs/code-quality-checks
This is just another technique to engineer quality outcomes; you’re just working from a different starting point.
The odds of success feel like gambling. 60%, or 40%, or worse. This is downstream of model quality.
Soon, 80%, 95%, 99%, 99.99%. Then, it won't be "gambling" anymore.
- One shot or "spray and pray" prompt only vibe coding: gambling.
- Spec driven TDD AI vibe coding: more akin to poker.
- Normal coding (maybe with tab auto complete): eating veggies/work.
Notably though gambling has the massive downside of losing your entire life and life savings. Being in the "vibe coding" bucket's worse case is being insufferable to your friends and family, wasting your time, and spending $200/month on a max plan.
Will Digbert be able to handle it or will he pretend to handle it? Or will he handle it in a way that it will break again in six weeks and will evolve into his full time job for a year?
If this is gambling, middle management has been gambling for too long.
I'm not arguing against accountability, only against gambling.
Opus specifically from 4.1 to 4.5 was such a major leap that some take it for granted, it went from getting stuck in loops, generally getting lost constantly, needing so so much attention to keep it going to being able to get a prompt, understand it from minimal context and produce what you wanted it to do. Opus 4.6 was a slight downgrade since it has issues with respecting what the user has to say.
Sometimes I can get away with 3K LoC PRs, sometimes I take a really long time on a +80 -25 change. You have to be intellectually honest with yourself about where to spend your time.
Which I assert is semantically equivalent to saying: Human drivers (even when operating at the diminished capacity of not even being present in the car) are less likely to make errors driving a car than AIs.
That is extremely stupid. What does that ban get you? I reqct to this because a friend mentioned exactly this. And I was dumbfounded.
CEO1: "We allow our engineers to use AI for all work."
CEO2: "Oh yea? We mandate our engineers use AI for at least N% of their work!"
CEO3: "You think that's good? We mandate our engineers use AI for all code!!"
CEO4: "Pfff, amateurs. We don't even allow our engineers to open source code editors or even look at the LLM output..."
confidence in firing coders I presume..
The results do speak for themselves, but it doesn't work.
Does it? It did in the past. Now it doesn't. Maybe "add a button to display a colour selector" really is the canonical way to code that feature, and the 100+ lines of generated code are just a machine language artifact like binary.
> But it robs me of the part that’s best for the soul. Figuring out how this works for me, finding the clever fix or conversion and getting it working. My job went from connecting these two things being the hard and reward part, to just mopping up how poorly they’ve been connected.
Skill issue. Two nights ago, I used Claude to write an iOS app to convert Live Photos into gifs. No other app does it well. I'm going to publish it as my first app. I wouldn't have bothered to do it without AI, and my soul feels a lot better with it.