My anxiety about falling behind with AI plummeted after I realized many of these tweets are overblown in this way. I use AI every day, how is everyone getting more spectacular results than me? Turns out: they exaggerate.
Here are several real stories I dug into:
"My brick-and-mortar business wouldn't even exist without AI" --> meant they used Claude to help them search for lawyers in their local area and summarize permits they needed
"I'm now doing the work of 10 product managers" --> actually meant they create draft PRD's. Did not mention firing 10 PMs
"I launched an entire product line this weekend" --> meant they created a website with a sign up, and it shows them a single javascript page, no customers
"I wrote a novel while I made coffee this morning" --> used a ChatGPT agent to make a messy mediocre PDF
Getting viral on X is the current replacement for selling courses for {daytrading,amazon FBA,crypto}.
The content of the tweets isn't the thing.. bull-posting or invoking Cunningham's Law is. X is the destination for formula posting and some of those blue checkmarks are getting "reach" rev share kickbacks.
Yeah, if you get enough impressions, you get some revenue, so you don't need to sell any courses, just viral content. Which is why some (not ALL) exaggerate as suggested.
It's a bit insane how much reach you need before you'd earn anything impactful, though.
I average 1-2M impressions/month, and have some video clips on X/Twitter that have gotten 100K+ views, and average earnings of around $42/month (over the past year).
I imagine you'd need hundreds of millions of impressions/views on Twitter to earn a living with their current rates.
At the end of the day, it doesn't really get you that much if you get 70% of the way there on your initial prompt (which you probably spent some time discussing, thinking through, clarifying requirements on). Paid, deliverable work is expected to involve validation, accountability, security, reliability, etc.
Taking that 70% solution and adding these things is harder than if a human got you 70% there, because the mistakes LLMs make are designed to look right, while being wrong in ways a sane human would never be. This makes their mistakes easy to overlook, requiring more careful line-by-line review in any domain where people are paying you. They also duplicate code and are super verbose, so they produce a ton tech debt -> more tokens for future agents to clog their contexts with.
I like using them, they have real value when used correctly, but I'm skeptical that this value is going to translate to massive real business value in the next few years, especially when you weigh that with the risk and tech debt that comes along with it.
> "I wrote a novel while I made coffee this morning" --> used a ChatGPT agent to make a messy mediocre PDF
There was a story years ago about someone who made hundreds of novels on Amazon, in aggregate they pulled in a decent penny. I wonder if someone's doing the same but with ChatGPT instead.
Pretty sure there was a whole era where people were doing this with public domain works, as well as works generated by Markov chains spitting out barely-plausible-at-first-glance spaghetti. I think that well started to dry up before LLMs even hit the scene.
Afaik, I think the way people are making money in this space is selling courses that teach you how to sell mass produced AI slop on Amazon, rather than actually doing it
“I used AI to make a super profitable stock trading bot”
—-> using fake money with historical data
“I used AI to make an entire NES emulator in an afternoon!” —-> a project that has been done hundreds of times and posted all over github with plenty of references
I actually read through the logs and the code in the rare instances someone actually posts their prompts and the generated output. If I'm being overly cynical about the tech, I want to know.
The last one I did it on was breathlessly touted as "I used [LLM] to do some advanced digital forensics!"
Dawg. The LLM grepped for a single keyword you gave it and then faffed about putting it into json several times before throwing it away and generating some markdown instead. When you told it the result was bad, it grepped for a second word and did the process again.
It looks impressive with all these json files and bash scripts flying by, but what it actually did was turn a single word grep into blog post markdown and you still had to help it.
Some of you have never been on enterprise software sales calls and it shows.
"I used AI to write a GPU-only MoE forward and backward pass to supplement the manual implementation in PyTorch that only supported a few specific GPUs" -> https://github.com/lostmsu/grouped_mm_bf16 100% vibe coded.
LLMs are amazing and I do seriously wonder if the singularity could happen in my lifetime ... but there definitely are people over-hyping present capabilities too much right now. If present models were fully human-level proper-extended-Turing-test-passing AGI, then the results on the economy and the software ecosystem would be as immediately obvious and world-changing as a comet impact.
I don't think Rakyll or Andrej are claiming these things; I think they're assuming their readers share more context with them and that it's not necessary to re-tread that every time they post about their surprise at AI currently being better than they expected. I've had the experience multiple times now of reading posts from people like them, nodding along, and then reading breathless quote-tweets of those very same posts exclaiming about how it means that AGI is here right now.
Being respected inside big companies has little to do with engagement on social media. Most of the best engineers are working hands-down. Arguably shitposting on the internet may have a negative correlation with technical ability inside Google.
One of the times I think the draconian approach Apple has towards employee speaking as an associate of the firm without explicit authorization is the correct one.
Perhaps nobody wants to have the uncomfortable conversation that AI is making the competent more competent and the incompetent less competent, because it would imply that AI provides brutally unequal benefits. The AI haters don't want this discussion because it would imply AI has any benefits, and the AI lovers don't want to have this discussion because it would imply the benefits of AI aren't universal and will increase inequality.
There are two major reasons people don't show proof about the impact of agentic coding:
1) The prompts/pipelines portain to proprietary IP that may or may not be allowed to be shown publically.
2) The prompts/pipelines are boring and/or embarrassing and showing them will dispel the myth that agentic coding is this mysterious magical process and open the people up to dunking.
For example in the case of #2, I recently published the prompts I used to create a terminal MIDI mixer (https://github.com/minimaxir/miditui/blob/main/agent_notes/P...) in the interest of transparency, but those prompts correctly indicate that I barely had an idea how MIDI mixing works and in hindsight I was surprised I didn't get harrassed for it. Given the contentious climate, I'm uncertain how often I will be open-sourcing my prompts going forward.
You weren't harassed for it because (1) it is interesting and (2) you were not hiding the AI involvement and passing it off as your own.
The results (for me) are very much hit-and-miss and I still see it as a means of last resort rather than a reliable tool that I know the up and downsides of. There is a pretty good chance you'll be wasting your time and every now and then it really moves the needle. It is examples like yours that actually help to properly place the tool amongst the other options.
> The prompts/pipelines are boring and/or embarrassing and showing them will dispel the myth that agentic coding is this mysterious magical process
You nailed it. Prompting is dull and self evident. Sure, you need basic skills to formulate a request. But it’s not a science and has nothing to do with engineering.
1) no one cares if it works. No one cared before how your code looked as long as you are not a known and well used opensource project.
2) there are plenty of services which do not require state or login and can't be hacked. So still plenty of use cases you can explore. But yes i do agree that Security for production live things are still the biggest worry. But lets be honest, if you do not have a real security person on your team, the shit outthere is not secure anyway. Small companies do not know how to build securely.
> 1) no one cares if it works. No one cared before how your code looked as long as you are not a known and well used opensource project.
Forgive me if this is overly blunt, but this is such a novice/junior mindset. There are many real world examples of things that "worked" but absolutely should not have, and when it blows up, can easily take out an entire company. Unprotected/unrestricted firebase keys living in the client are all the rage right now, yea they "work"until someone notices "hey, I technically have read/write god mode access to their entire prod DB", and then all of a sudden it definitely doesn't work and you've possibly opened yourself to a huge array of legal problems.
The more regulated the industry and the more sensitive the business data, the worse this is exacerbated. Even worse if you're completely oblivious to the possibility of these kinds of things.
> Forgive me if this is overly blunt, but this is such a novice/junior mindset.
Unfortunately the reality is there are far more applications written (not just today but for many years now) by developer teams that will include a dozen dependencies with zero code review because feature XYZ will get done in a few days instead of a few weeks.
And yes, that often comes back to bite the team (mostly in terms of maintenance burden down the road, leading to another full rebuild), but it usually doesn't affect the programmers who are making the decisions, or the project managers who ship the first version.
You seem to think I'm an AI coding hater or something. I'm not. I think these tools are incredibly useful and I use them daily. However, like described in the article, I do am skeptical about stories where AI writes whole applications, SaaS or game engines in a few hours and everything "just works". That is not my experience.
The Cloudflare OAuth lib is impressive, I will readily admit that. But they also clearly mention that of course everything was carefully reviewed, and that not everything was perfect but that the AI was mostly able to fix things when told to. This was surely still a lot of work, which makes this story also much more realistic in my opinion. It surely greatly sped up the process of writing an OAuth library - how much exactly is however hard to say. Especially in security-relevant code, the review process is often longer than the actual writing of the code.
I'm fundamentally a hobbyist programmer, so I would have no problem sharing my process.
However, I'm not nearly organized enough to save all my prompts! I've tried to do it a few times for my own reference. The thing is, when I use Claude Code, I do a lot of:
- Going back and revising a part of the conversation and trying again—sometimes reverting the code changes, sometimes not.
- Stopping Claude partway through a change so I can make manual edits before I let Claude continue.
- Jumping between entirely different conversation histories with different context.
And so on. I could meticulously document every action, but it quickly gets in the way of experimentation. It's not entirely different from trying to write down every intermediate change you make in your code editor, between actual VCS commits.
I guess I could record my screen, but (A) I promise you don't actually want to watch me fiddle with Claude for hours and (B) it would make me too self-conscious.
It would be very cool to have a tool that goes through Claude's logs and exports some kind of timeline in a human-readable format, but I would need it to be automated.
---
Also, if you can't tell from the above, my use of Claude is very far from "type a prompt, get a finished program." I do a lot of work in order to get useful output. I happen to really enjoy coding this way, and I've gotten great results, but it's not like I'm entering a prompt and then taking a nap.
Could you clarify that last paragraph for me? I’m not sure what ”contentious climate” is here. AI antihype? I don’t understand the connection to not being harassed for something, isn’t that a good thing rather than something that would make you uncertain if you want to share prompts in the future?
"AI tech bro creates slop X because they don't understand how X actually works" is a common trope among the anti-AI crowd even on Hacker News that has only been increasing in recent months, and sharing prompts/pipelines provides strong evidence that can be pointed at for dunks. Sharing AI workflows is more likely to illicit this snark if the project breaks out of the AI bubble, though in the case of the AI boosters on X described as in the HN submission that's a feature due to how monetization works that platform. It's not something I want to encourage for my own projects, though.
There's also the lessons on the recent shitstorms in the gaming industry, with Sandfall about Expedition 33's use of GenAI and Larian's comments on GenAI with concept art, where both received massive backlash because they were transparent in interviews about how GenAI was (inconsequentially) used. The most likely consequence of those incidents is that game developers are less likely on their development pipelines.
you can use however you like, no one cares. really, no one.
but, people in general are NOT inclined to pay for AI slop. that is the controversy.
why would I waste my time reading garbage words generated by an LLM? If people wanted this, they would go to the llm themselves.
the whole point of artistic expression is to present oneself, to share a perspective. llms do not have a singular point of view, they do not have a perspective, they do not have an cohesive aggregate of experiences. they just regurgitate the average form.
no one is interested in this. even when distributed for free, is disrespectful to others that put their time until they realized is just hot garbage yet again.
people are getting tired of low effort `content`, yet again another unity or unreal engine resking, asset flipping `game`...
you get the idea, lots of people will feel offended and disrespected when presented with no effort. got it? it is not exclusively about intellectual property theft also, i don't care about it, i just hate slop.
now whether you like it or not, the new meta is to not look professional. the more personal, the better.
AI is cool for a lot of things, searching, learning, natural language apropos, profiling, surveilling, compressing information...it is fantastic technology!
not a replacement for art, never will be.
You'd think so, but with the recent extreme polarization of GenAI the common argument among the anti-AI crowd is the absolute "if AI touched it, it's slop". For example in the Expedition 33 case (which won Game of the Year), even though the GenAI asset was clearly a placeholder and replaced 2 days after launch, a surprisingly large number of players said sincerely "I enjoyed my time with E33 but after finding out they used GenAI I no longer enjoy it."
In a lesser example, a week ago a Rust developer on Bluesky tried to set up a "Tainted Slopware" list of OSS which used AI, but the criteria for inclusion was as simple as "they accepted an AI-generated PR" and "the software can set up a MCP server." It received some traction but eventually imploded, partially due to the fact that the Linux kernel would be considered slopware due to that criteria.
Sure, but I'm gonna push back and go "so what"? That sort of thing is what haters do, especially in the notoriously toxic world of gaming.
"Some people expressed disappointment about a thing I think is silly" is literally the center square on the gamer outrage bingo card lol. Same with "someone made a list that I think is kind of stupid".
And again, so what? Why should you care? Again, if you feel that insecure about it, it's you and your work that's the problem, not the haters who are always going to exist. Have the courage of your own convictions or maybe admit that it isn't that strong of a conviction lol.
Doesn't the existence of consumer products like ChatGPT indicate that LLMs aren't able to do human-level work? If OpenAI really had a digital workforce with the capabilities of ~100k programmers/scientists/writers/lawyers/doctors etc, wouldn't the most profitable move be to utilize those "workers" directly, rather that renting out their skills piecemeal?
If this stuff worked, then given that Microsoft has a huge share of it, shouldn't Microsoft's products be good? Or at least getting better. I use Bing most days [1], and it's consistently an absolute joke.
[1] to farm reward points to get cosmetic items in video games
That depends on what the real value is. The sure way to get rich selling pickaxes to gold miners. However you would be even richer if you figured out where the gold really was and mined in that exact location.'
> Antigravity did the vast amount of work, it feels unworthy
I think this is true for me as well. I have two types of projects that I’ve been working on - small ones with a mix of code I wrote and AI. I have posted these, as I spent a lot of time guiding the AI, cleaning up the AI’s output, and I think the overall project has value that others can learn from and built on.
But I also have some that are almost 100% vibe-coded. First, those would take a lot of time to clean up and write documentation for to make them publishable/useful.
But also, I do think they feel “unworthy”. Even though I think they can be helpful, and I was looking for open-source versions of those things. But how valuable can it really be if I was able to vibe-code it in a few prompts? The next person looking for it will probably do the same thing I did and vibe-code their own version after a few minutes.
I think this
"* The software is pretty specific to my requirements."
is the biggest part for me. I built something with Antigravity over the holidays that I can use for myself and it solves my use case.
I tried thinking about if this can be helpful for others and pushed it a bit further into a version that could be hosted. Which does not make that much sense because it is a computationally intense numerical solver for thermal bridges and just awfully slow on a free hosted platform.
But the project was a couple of evenings and would otherwise haven taken me half a year to complete (and thus never been done).
This is a strage phenomenon where people get excited by the mere fact that someone else is excited by something which is not directly visible to the spectator. It works well in horror movies and as it seems with AI hype.
Writing code is the easiest thing to do at Google. Getting past layers of hierarchy and nailing down what the code will actually do and who gets credit for it will take years for a major project.
I’ve taken to calling this (in my mind) the Age of the Sycophants. In politics, in corporate life, in technology and in social media, many people are building a public life around saying things that others want to hear, with demonstrably zero relationship to truth or even credibility.
I'm still waiting for all these remarkable achievements produced with this new technology to provide tangible value to the world. Surely we should be seeing groundbreaking products and innovations in all industries by now, improving the lives of millions of people.
Instead all we get is anecdata from influencers and entrepreneurs, and the technology being shoved into every brand and product. It's exhausting.
At least it seems that the tide is starting to turn. Perhaps we are at the downward slope of the Peak of Inflated Expectations.
Good article. "hype first and context later". Loads of online content has become highly sentational. I notice this on Youtube (especially thumnails and titles) Seems to be a trend. I wonder if -- collectively -- we'll develop a "shield" for this: (more critical thinking?)
If you don't get the results you don't get the results. If someone else can use this tool to get the results, they'll out-compete you. If they can't, then they've wasted time and you'll out-compete them. I see these influencer guys as idea-generators. It's super-cheap to test out some of these theories: e.g. how well Claude can do 3D modeling was an idea I wanted to test and I did and it's pretty good; I wanted to test Claude as a debugging aid and it's a huge help for me.
But I would never sit down to convince a person who is not a friend. If someone wanted me to do that, I'd expect to charge them for it. So the guys who are doing it for free are either peddling bullshit or they have some other unspecified objective and no one likes that.
Yes! And like … “out-compete” assumes a zero sum game. There are massive industries where the tools used to serve different market segments only barely overlap decades after the “game changing” tool was made available.
Like, screw the whole “artisanal small-batch software” argument—there are massive new systems being written in C every day despite decades of claims that it is an obsolete language doomed to be replaced by better alternatives. Those alternatives caught on in some segments and not in others. Electric cars caught on in some segments and not in others. Steel-and-concrete building construction caught on in some segments and not in others. It’ll be fine.
Great post! Indeed, it s deeply disappointing to see how both the tech industry and scientific community have fallen into the same attention-seeking trap: hyping their work with vague, sensational claims, only to later "clarify" with far more grounded—and often mundane—statements.
This tactic mirrors the strategies of tabloids, demagogues, and social media’s for-profit engagement playbook (think Zuckerberg, Musk, and the like). It’s a race to the bottom, eroding public trust and undermining the foundations of our society - all for short-term personal gain.
What’s even more disheartening is how this dynamic rewards self-promotion over substance. Today’s "experts" are often those who excel at marketing themselves, while the most knowledgeable and honest voices remain in the shadows. Their "flaw"? Refusing to sacrifice integrity for attention.
Anecdotally, I’m finding that, at least in the Spark ecosystem, AI-generated ideas and code are far from optimal. Some of this comes from misinterpreting the (sometimes poor) documentation, and some of it comes from, probably, there not being as many open source examples as CRUD apps, which AI “influentists” (to borrow from TFA) appear to often be hyping up.
This matters a lot to us because the difference in performance of our workflows can be the difference in $10/day in costs and $1000/day in costs.
Just like TFA stresses, it’s the expertise in the team that pushes back against poor AI-generated ideas and code that is keeping our business within reach of cash flow positive. ~”Surely this isn’t the right way to do this?”
The article nails the pattern but I think it's fundamnetally an incentives problem.
We're drowning in tweets, posts, news... (way more than anyone can reasonably consume). So what rises to the top? The most dramatic, attention-grabbing claims. "I built in 1 hour what took a team months" gets 10k retweets. "I used AI to speed up a well-scoped prototype after weeks of architectural thinking" gets...crickets
Social platforms are optimized for engagement, not accuracy. The clarification thread will always get a fraction of the reach of the original hype. And the people posting know this.
The frustrating part is there's no easy fix. Calling it out (like this article does) get almost no attention. And the nuanced followup never catches up with the viral tweet.
AI is like flossing. You waste more time listening to other people's opinions on whether it's helpful, than just trying it out yourself for a few days.
I never read the tweet as anything other than that an expert with deep knowledge of their domain was able to produce a PoC. Which I still find to be very exciting and worthy of being promoted. This article didn't really debunk much.
these are the kinds of people that can use generative AI best IMO. Deep domain knowledge is needed to spot when the model output is wrong even though it sounds 100% correct. I've seen people take a model's output as correct to a shocking degree like placing large bets at a horse track after uploading a pic of the schedule to ChatGPT. Many people believe whatever a computer tells them but, in their defense, no one has had to question a large calculation done by a calculator until now.
Totally agree. The more you know about software development the better you can use the tools. A noob trying to write a kernel driver would quickly hit a brick wall and give up. It's kind of self correcting.
Like everything in LLM land it's all about the prompt and agent pipeline. As others say below, these people are experts in their domain. Their prompts are essentially a form of codifying their own knowledge, as in Rakyll and Galen's examples, to achieve specific outcomes based on years and maybe even decades of work in the problem domain. It's no surprise their outputs when ingested by an LLM are useful, but it might not tell us much about the true native capability of a given AI system.
Almost every aspect of public life on social media nowadays is guided by sensationalism. It's simply a numbers game, and the "number" is engagement. Why would you do anything that's not completely geared towards engagement?
Its a strange phenomenon. You want to call out the bs but then you are just giving them engagement and boost. You want to stay away but there is a sort of confluence where these guys tend to ride on each others' post and boosts those posts anyway. If you ask questions, very rarely they answer, and if they do, it takes one question to unearth that it was the prompt or the skill. Eg: huggingface people post about claude finetuning models. how? when they gave everything in a skill file, and claude knew what scripts to write. Tinker is trying the same strategy. (yes, its impressive that claude could finetune, but not as impressive as the original claim that made me pay attention to the post)
It does not matter if they get the details wrong, its just that it needs to be vague enough, and exciting enough. Infact vagueness and not sharing the code part signals they are doing something important or they are 'in the know' which they cannot share. The incentives are totally inverted.
Great article. This needs to be framed. The whole trust me bro, and shock and awe of social medias is a non-stop assault these days. You can't open a wall without seeing those promoted up front and centre and without any proof.
If AI was so good today, why isn't there an explosion of successful products? All we see is these half baked "zomg so good bro!" examples that are technically impressive, but decisively incomplete or really, proof of concepts.
I'm not saying LLMs aren't useful, but they're currently completely misrepresented.
Hype sells clicks, not value. But, whatever floats the investors' boat...
A recent favorite of mine is simonw who usually is unable to stop hyping LLMs suddenly forgetting they exist in order to rhetorically "win" an argument:
> If you're confident that you know how to securely configure and use Wireguard across multiple devices then great
This is very well said - it is much nicer and more professional than the sentiments I could express on the matter.
The age of niche tech microcelebrities is on us. It existed a bit in the past (ESR, Uncle Bob, etc), but is much more of a thing now. Some of them make great content and don't say ridiculous things. Others not so much.
Idk man, all AI discussion feels like a waste of effort.
“yes it will”, “no it won’t” - nobody really knows, it's just a bunch of extremely opinionated people rehashing the same tired arguments across 800 comments per thread.
There’s no point in talking about it anymore, just wait to see how it all turns out.
The cool thing is you can just try it. The barrier to entry is incredibly low right now. If it works for you, great. If it doesn't work for you, great. At least you know.
That's the real reason the conversation seems pointless. Every thread is full of comments from one group saying how useful AI is, and from another group saying how useless it is. The first group is convinced the second group just hasn't figured out how to use it right, and the second group is convinced the first group is deluded or outright lying.
It's not "yes it is" vs "no it won't" though. The discussion is "Yes it does" vs "no it doesn't" (present tense.) There's nothing wrong with guessing about the future, but lying about a present that is testable and unwillingness to submit to the testing is wrong.
Even then nothing is learned. Every HN thread there is on AI coding: "I am using $model for writing software and its great." "I am using $model for writing software and it sucks and will never do it." 800 comments of that tit for tat in present tense. Still nothing learned.
Doesn't help that no one talks about exactly what they are doing and exactly how they are doing it, because capitalism vs open technology discussions meant to uplift the species.
Its still not a Hype, its still crazy what is possible today and we still have no clear at all if this progress continues as it does or not with the implication, that if it continues, it has major implications.
My wife, who has no clue about coding at all, chatgpted a very basic android app only with guidance of chatgpt. She would never ever been able to do this in 5 hours or so without my guidance. I DID NOT HELP HER at all.
I'm 'vibecoding' stuff small stuff for sure, non critical things for sure but lets be honest, i'm transforming a handfull of sentences and requirements into real working code, today.
Gemini 3 and Claude Opus 4.5 def feel better than their prevous versions.
Do they still fail? Yeah for sure but thats not the point.
The industry continues to progress on every single aspect of this: Tooling like claude CLI, Gemini CLI, Intellij integration, etc., Context length, compute, inferencing time, quality, depth of thinking etc. there is no current plateau visible at all.
And its not just LLMs, its the whole ecosystem of Machine Learning stuff: Highhly efficient weather model from google, Alpha fold, AlphaZero, Roboticsmovement, Environment detection, Image segmentation, ...
And the power of claude for example, you will only get with learning how to use it. Like telling it your coding style, your expectations regarding tests etc. We often assume, that an LLM should just be the magic work collegue 10x programmer but its everything an dnothing. If you don't communicate well enough it is not helpful.
And LLMs are not just good in coding, its great in reformulating emails, analysing error messages, writing basic SVG files, explaining kubernetes cluster status, being a friend for some people (see character.ai), explaining research paper, finding research, summarizing text, the list is way to long.
Alone 2026 there will go so many new datacenters live which will add so much more compute again, that the research will continue to be faster and more efficient.
There is also no current bubble to burst, Google fights against Microsoft, Antrophic and co. while on a global level USA competets with China and the EU on this technology. The richest companies on the planet are investing in this tech and they did not do this with bitcoins because they understod that bitcoin is stupid. But AI is not stupid.
Or Machine learing is not stupid.
Do not underestimate the current status of AI tools we have, do not underestimate the speed, continues progress and potential exponential growth of this.
My timespan expecation for obvious advancments in AI is 5-15 years. Experts in this field predict already 2027/2030.
But to iterate over this: a few years ago no one would have had a good idea how we could transform basic text into complex code in such a robust way, which such diverse input (different language, missing specs, ...) . No one. Even 'just generating a website'.
> My wife, who has no clue about coding at all, chatgpted a very basic android app only with guidance of chatgpt. She would never ever been able to do this in 5 hours or so without my guidance. I DID NOT HELP HER at all.
Literally all much of the business world needed was a slightly more capable VB but now we have a bunch of crappy, web browser based SaaS platforms instead.
I think it really depends how a person judges the progress from chatgpt 3.5, 3 years ago to Opus 4.5.
In one light it is super impressive and amazing progress, in another light it is not impressive at all and totally over hyped.
Using the Hubert Dreyfus analogy. It is impressive if the goal is to climb as high as we can up giant tree. The height we have reached isn't impressive at all though if we are trying to climb the tree to get to the moon.
Even if we assume for a moment everything you are saying is true and/or reasonable, can't you see how comments like these paint your position here in a bad light? It just reads a little desperate!
It might be just different viewpoints people don't understand?
I'm advocating for spending time with AI because it works already good enough and it continues to progress surprisingly fast. Unexperienced fast for me tbh.
If i say "AI is great" i also know when AI is also stupid but i'm already/stil so impressed that i can transform basic text into working go/java whatever code, that i accept that its not perfect just because I highly appreciate how fast we got this.
And it feels weird too tbh. It doesn't feel special to talk to an LLM and get code back somehow while this was unthinkable just a few years back.
Somethimes it likes you just forget about all these facts and have to remind yourself that this is something new.
Just because it’s “new” doesn’t mean it’s going to fulfill all our wildest fantasies. It’s becoming clear that these things are just… tools. Useful in certain situations, but ultimately not the game-changer they’re sold as.
Here are several real stories I dug into:
"My brick-and-mortar business wouldn't even exist without AI" --> meant they used Claude to help them search for lawyers in their local area and summarize permits they needed
"I'm now doing the work of 10 product managers" --> actually meant they create draft PRD's. Did not mention firing 10 PMs
"I launched an entire product line this weekend" --> meant they created a website with a sign up, and it shows them a single javascript page, no customers
"I wrote a novel while I made coffee this morning" --> used a ChatGPT agent to make a messy mediocre PDF
The content of the tweets isn't the thing.. bull-posting or invoking Cunningham's Law is. X is the destination for formula posting and some of those blue checkmarks are getting "reach" rev share kickbacks.
If it was successful, they wouldnt be telling everyone about it
I average 1-2M impressions/month, and have some video clips on X/Twitter that have gotten 100K+ views, and average earnings of around $42/month (over the past year).
I imagine you'd need hundreds of millions of impressions/views on Twitter to earn a living with their current rates.
Taking that 70% solution and adding these things is harder than if a human got you 70% there, because the mistakes LLMs make are designed to look right, while being wrong in ways a sane human would never be. This makes their mistakes easy to overlook, requiring more careful line-by-line review in any domain where people are paying you. They also duplicate code and are super verbose, so they produce a ton tech debt -> more tokens for future agents to clog their contexts with.
I like using them, they have real value when used correctly, but I'm skeptical that this value is going to translate to massive real business value in the next few years, especially when you weigh that with the risk and tech debt that comes along with it.
There was a story years ago about someone who made hundreds of novels on Amazon, in aggregate they pulled in a decent penny. I wonder if someone's doing the same but with ChatGPT instead.
“I used AI to make an entire NES emulator in an afternoon!” —-> a project that has been done hundreds of times and posted all over github with plenty of references
It is really good in doing this.
those ideas are like UI experiments or small tools helping me doing stuff.
Its also super great in ELI5'ing anything
The last one I did it on was breathlessly touted as "I used [LLM] to do some advanced digital forensics!"
Dawg. The LLM grepped for a single keyword you gave it and then faffed about putting it into json several times before throwing it away and generating some markdown instead. When you told it the result was bad, it grepped for a second word and did the process again.
It looks impressive with all these json files and bash scripts flying by, but what it actually did was turn a single word grep into blog post markdown and you still had to help it.
Some of you have never been on enterprise software sales calls and it shows.
Welcome to the internet
I don't think Rakyll or Andrej are claiming these things; I think they're assuming their readers share more context with them and that it's not necessary to re-tread that every time they post about their surprise at AI currently being better than they expected. I've had the experience multiple times now of reading posts from people like them, nodding along, and then reading breathless quote-tweets of those very same posts exclaiming about how it means that AGI is here right now.
One of the times I think the draconian approach Apple has towards employee speaking as an associate of the firm without explicit authorization is the correct one.
Google engineer says Claude Code built in one hour what her team spent a year on - https://news.ycombinator.com/item?id=46477966 - Jan 2026 (81 comments)
1) The prompts/pipelines portain to proprietary IP that may or may not be allowed to be shown publically.
2) The prompts/pipelines are boring and/or embarrassing and showing them will dispel the myth that agentic coding is this mysterious magical process and open the people up to dunking.
For example in the case of #2, I recently published the prompts I used to create a terminal MIDI mixer (https://github.com/minimaxir/miditui/blob/main/agent_notes/P...) in the interest of transparency, but those prompts correctly indicate that I barely had an idea how MIDI mixing works and in hindsight I was surprised I didn't get harrassed for it. Given the contentious climate, I'm uncertain how often I will be open-sourcing my prompts going forward.
The results (for me) are very much hit-and-miss and I still see it as a means of last resort rather than a reliable tool that I know the up and downsides of. There is a pretty good chance you'll be wasting your time and every now and then it really moves the needle. It is examples like yours that actually help to properly place the tool amongst the other options.
You nailed it. Prompting is dull and self evident. Sure, you need basic skills to formulate a request. But it’s not a science and has nothing to do with engineering.
1) the code AI produces is full of problems, and if you show it, people will make fun of you, or
2) if you actually run the code as a service people can use, you'll immediately get hacked by people to prove that the code is full of problems.
2) there are plenty of services which do not require state or login and can't be hacked. So still plenty of use cases you can explore. But yes i do agree that Security for production live things are still the biggest worry. But lets be honest, if you do not have a real security person on your team, the shit outthere is not secure anyway. Small companies do not know how to build securely.
Forgive me if this is overly blunt, but this is such a novice/junior mindset. There are many real world examples of things that "worked" but absolutely should not have, and when it blows up, can easily take out an entire company. Unprotected/unrestricted firebase keys living in the client are all the rage right now, yea they "work"until someone notices "hey, I technically have read/write god mode access to their entire prod DB", and then all of a sudden it definitely doesn't work and you've possibly opened yourself to a huge array of legal problems.
The more regulated the industry and the more sensitive the business data, the worse this is exacerbated. Even worse if you're completely oblivious to the possibility of these kinds of things.
Unfortunately the reality is there are far more applications written (not just today but for many years now) by developer teams that will include a dozen dependencies with zero code review because feature XYZ will get done in a few days instead of a few weeks.
And yes, that often comes back to bite the team (mostly in terms of maintenance burden down the road, leading to another full rebuild), but it usually doesn't affect the programmers who are making the decisions, or the project managers who ship the first version.
I have seen production databases reachable from the internet with 8 character password and plenty others.
But my particular point is only about the readability of code from others.
The Cloudflare OAuth lib is impressive, I will readily admit that. But they also clearly mention that of course everything was carefully reviewed, and that not everything was perfect but that the AI was mostly able to fix things when told to. This was surely still a lot of work, which makes this story also much more realistic in my opinion. It surely greatly sped up the process of writing an OAuth library - how much exactly is however hard to say. Especially in security-relevant code, the review process is often longer than the actual writing of the code.
However, I'm not nearly organized enough to save all my prompts! I've tried to do it a few times for my own reference. The thing is, when I use Claude Code, I do a lot of:
- Going back and revising a part of the conversation and trying again—sometimes reverting the code changes, sometimes not.
- Stopping Claude partway through a change so I can make manual edits before I let Claude continue.
- Jumping between entirely different conversation histories with different context.
And so on. I could meticulously document every action, but it quickly gets in the way of experimentation. It's not entirely different from trying to write down every intermediate change you make in your code editor, between actual VCS commits.
I guess I could record my screen, but (A) I promise you don't actually want to watch me fiddle with Claude for hours and (B) it would make me too self-conscious.
It would be very cool to have a tool that goes through Claude's logs and exports some kind of timeline in a human-readable format, but I would need it to be automated.
---
Also, if you can't tell from the above, my use of Claude is very far from "type a prompt, get a finished program." I do a lot of work in order to get useful output. I happen to really enjoy coding this way, and I've gotten great results, but it's not like I'm entering a prompt and then taking a nap.
There's also the lessons on the recent shitstorms in the gaming industry, with Sandfall about Expedition 33's use of GenAI and Larian's comments on GenAI with concept art, where both received massive backlash because they were transparent in interviews about how GenAI was (inconsequentially) used. The most likely consequence of those incidents is that game developers are less likely on their development pipelines.
but, people in general are NOT inclined to pay for AI slop. that is the controversy.
why would I waste my time reading garbage words generated by an LLM? If people wanted this, they would go to the llm themselves. the whole point of artistic expression is to present oneself, to share a perspective. llms do not have a singular point of view, they do not have a perspective, they do not have an cohesive aggregate of experiences. they just regurgitate the average form. no one is interested in this. even when distributed for free, is disrespectful to others that put their time until they realized is just hot garbage yet again.
people are getting tired of low effort `content`, yet again another unity or unreal engine resking, asset flipping `game`...
you get the idea, lots of people will feel offended and disrespected when presented with no effort. got it? it is not exclusively about intellectual property theft also, i don't care about it, i just hate slop.
now whether you like it or not, the new meta is to not look professional. the more personal, the better.
AI is cool for a lot of things, searching, learning, natural language apropos, profiling, surveilling, compressing information...it is fantastic technology! not a replacement for art, never will be.
If your hand is good, throw it down and let the haters weep. If you scared to show your cards, you don't have a good hand and you're bluffing.
In a lesser example, a week ago a Rust developer on Bluesky tried to set up a "Tainted Slopware" list of OSS which used AI, but the criteria for inclusion was as simple as "they accepted an AI-generated PR" and "the software can set up a MCP server." It received some traction but eventually imploded, partially due to the fact that the Linux kernel would be considered slopware due to that criteria.
"Some people expressed disappointment about a thing I think is silly" is literally the center square on the gamer outrage bingo card lol. Same with "someone made a list that I think is kind of stupid".
And again, so what? Why should you care? Again, if you feel that insecure about it, it's you and your work that's the problem, not the haters who are always going to exist. Have the courage of your own convictions or maybe admit that it isn't that strong of a conviction lol.
[1] to farm reward points to get cosmetic items in video games
Of course you can also get rich selling scams.
That said, I use Antigravity with great success for self hosted software. I should publish it.
Why haven't I?
* The software is pretty specific to my requirements.
* Antigravity did the vast amount of work, it feels unworthy?
* I don't really want a project, but that shouldn't really stop me pushing to a public repo.
* I'm a bit hesitant to "out" myself?
Nonetheless, even though I'm not the person, I'm surprised there isn't more evidence out there.
I think this is true for me as well. I have two types of projects that I’ve been working on - small ones with a mix of code I wrote and AI. I have posted these, as I spent a lot of time guiding the AI, cleaning up the AI’s output, and I think the overall project has value that others can learn from and built on.
But I also have some that are almost 100% vibe-coded. First, those would take a lot of time to clean up and write documentation for to make them publishable/useful.
But also, I do think they feel “unworthy”. Even though I think they can be helpful, and I was looking for open-source versions of those things. But how valuable can it really be if I was able to vibe-code it in a few prompts? The next person looking for it will probably do the same thing I did and vibe-code their own version after a few minutes.
https://github.com/schoenenbach/thermal-bridge https://thermal-bridge.streamlit.app/
Instead all we get is anecdata from influencers and entrepreneurs, and the technology being shoved into every brand and product. It's exhausting.
At least it seems that the tide is starting to turn. Perhaps we are at the downward slope of the Peak of Inflated Expectations.
But I would never sit down to convince a person who is not a friend. If someone wanted me to do that, I'd expect to charge them for it. So the guys who are doing it for free are either peddling bullshit or they have some other unspecified objective and no one likes that.
Like, screw the whole “artisanal small-batch software” argument—there are massive new systems being written in C every day despite decades of claims that it is an obsolete language doomed to be replaced by better alternatives. Those alternatives caught on in some segments and not in others. Electric cars caught on in some segments and not in others. Steel-and-concrete building construction caught on in some segments and not in others. It’ll be fine.
This tactic mirrors the strategies of tabloids, demagogues, and social media’s for-profit engagement playbook (think Zuckerberg, Musk, and the like). It’s a race to the bottom, eroding public trust and undermining the foundations of our society - all for short-term personal gain.
What’s even more disheartening is how this dynamic rewards self-promotion over substance. Today’s "experts" are often those who excel at marketing themselves, while the most knowledgeable and honest voices remain in the shadows. Their "flaw"? Refusing to sacrifice integrity for attention.
This matters a lot to us because the difference in performance of our workflows can be the difference in $10/day in costs and $1000/day in costs.
Just like TFA stresses, it’s the expertise in the team that pushes back against poor AI-generated ideas and code that is keeping our business within reach of cash flow positive. ~”Surely this isn’t the right way to do this?”
- accountability
- reliability
- validation
- security
- liability
Humans can reliably produce text with all of these features. LLMs can reliably produce text with none of them.
If it doesn't have all of these, it could still be worth paying for if it's novel and entertaining. IMO, LLMs can't really do that either.
We're drowning in tweets, posts, news... (way more than anyone can reasonably consume). So what rises to the top? The most dramatic, attention-grabbing claims. "I built in 1 hour what took a team months" gets 10k retweets. "I used AI to speed up a well-scoped prototype after weeks of architectural thinking" gets...crickets
Social platforms are optimized for engagement, not accuracy. The clarification thread will always get a fraction of the reach of the original hype. And the people posting know this.
The frustrating part is there's no easy fix. Calling it out (like this article does) get almost no attention. And the nuanced followup never catches up with the viral tweet.
these are the kinds of people that can use generative AI best IMO. Deep domain knowledge is needed to spot when the model output is wrong even though it sounds 100% correct. I've seen people take a model's output as correct to a shocking degree like placing large bets at a horse track after uploading a pic of the schedule to ChatGPT. Many people believe whatever a computer tells them but, in their defense, no one has had to question a large calculation done by a calculator until now.
It does not matter if they get the details wrong, its just that it needs to be vague enough, and exciting enough. Infact vagueness and not sharing the code part signals they are doing something important or they are 'in the know' which they cannot share. The incentives are totally inverted.
> The tech community must shift its admiration back toward reproducible results and away from this “trust-me-bro” culture.
Well said, in my opinion.
Sitting 2 hours with an Ai agent developing end to end products does.
If AI was so good today, why isn't there an explosion of successful products? All we see is these half baked "zomg so good bro!" examples that are technically impressive, but decisively incomplete or really, proof of concepts.
I'm not saying LLMs aren't useful, but they're currently completely misrepresented.
Hype sells clicks, not value. But, whatever floats the investors' boat...
> If you're confident that you know how to securely configure and use Wireguard across multiple devices then great
https://news.ycombinator.com/item?id=46581183
What happened to your overconfidence in LLMs ability to help people without previous experience do something they were unable to before?
Masks during Covid and LLMs, used as political pawns. It’s kind of sad.
The age of niche tech microcelebrities is on us. It existed a bit in the past (ESR, Uncle Bob, etc), but is much more of a thing now. Some of them make great content and don't say ridiculous things. Others not so much.
Even tech executives are aping it...
Someone mentioned to me they're like the historical paper boys who used to yell Extra Extra and announcing something trying to sell newspapers.
“yes it will”, “no it won’t” - nobody really knows, it's just a bunch of extremely opinionated people rehashing the same tired arguments across 800 comments per thread.
There’s no point in talking about it anymore, just wait to see how it all turns out.
Doesn't help that no one talks about exactly what they are doing and exactly how they are doing it, because capitalism vs open technology discussions meant to uplift the species.
My wife, who has no clue about coding at all, chatgpted a very basic android app only with guidance of chatgpt. She would never ever been able to do this in 5 hours or so without my guidance. I DID NOT HELP HER at all.
I'm 'vibecoding' stuff small stuff for sure, non critical things for sure but lets be honest, i'm transforming a handfull of sentences and requirements into real working code, today.
Gemini 3 and Claude Opus 4.5 def feel better than their prevous versions.
Do they still fail? Yeah for sure but thats not the point.
The industry continues to progress on every single aspect of this: Tooling like claude CLI, Gemini CLI, Intellij integration, etc., Context length, compute, inferencing time, quality, depth of thinking etc. there is no current plateau visible at all.
And its not just LLMs, its the whole ecosystem of Machine Learning stuff: Highhly efficient weather model from google, Alpha fold, AlphaZero, Roboticsmovement, Environment detection, Image segmentation, ...
And the power of claude for example, you will only get with learning how to use it. Like telling it your coding style, your expectations regarding tests etc. We often assume, that an LLM should just be the magic work collegue 10x programmer but its everything an dnothing. If you don't communicate well enough it is not helpful.
And LLMs are not just good in coding, its great in reformulating emails, analysing error messages, writing basic SVG files, explaining kubernetes cluster status, being a friend for some people (see character.ai), explaining research paper, finding research, summarizing text, the list is way to long.
Alone 2026 there will go so many new datacenters live which will add so much more compute again, that the research will continue to be faster and more efficient.
There is also no current bubble to burst, Google fights against Microsoft, Antrophic and co. while on a global level USA competets with China and the EU on this technology. The richest companies on the planet are investing in this tech and they did not do this with bitcoins because they understod that bitcoin is stupid. But AI is not stupid.
Or Machine learing is not stupid.
Do not underestimate the current status of AI tools we have, do not underestimate the speed, continues progress and potential exponential growth of this.
My timespan expecation for obvious advancments in AI is 5-15 years. Experts in this field predict already 2027/2030.
But to iterate over this: a few years ago no one would have had a good idea how we could transform basic text into complex code in such a robust way, which such diverse input (different language, missing specs, ...) . No one. Even 'just generating a website'.
you know Google used to have a app for this
https://www.youtube.com/watch?v=8ADwPLSFeY8
I swear people have forgotten how productive native programming 30 years ago was (Delphi, even VB)
compared to the disaster that is the web today
When that hole wasn't as big, it seemed fewer software projects failed.
In one light it is super impressive and amazing progress, in another light it is not impressive at all and totally over hyped.
Using the Hubert Dreyfus analogy. It is impressive if the goal is to climb as high as we can up giant tree. The height we have reached isn't impressive at all though if we are trying to climb the tree to get to the moon.
I'm advocating for spending time with AI because it works already good enough and it continues to progress surprisingly fast. Unexperienced fast for me tbh.
If i say "AI is great" i also know when AI is also stupid but i'm already/stil so impressed that i can transform basic text into working go/java whatever code, that i accept that its not perfect just because I highly appreciate how fast we got this.
And it feels weird too tbh. It doesn't feel special to talk to an LLM and get code back somehow while this was unthinkable just a few years back.
Somethimes it likes you just forget about all these facts and have to remind yourself that this is something new.
This is truly impressive and not only hype.
Things have been impressive at least since April 2025.
> Your Firefox settings blocked this content from tracking you across sites or being used for ads.
Why is this website serving such crap?
For God's sake, if there is anything absolutely worth showing on X, just include a screenshot of it instead of subjecting us all to that malware.