AI is a big fat lie!

Discussion
1

I want to share my experience with AI agents enough to understand how they work in the real world!

I wanted to create a new slider with a clear idea in mind adn I built most of it until I decided to give control to AI. I got lazy, and boy, I am sorry now. The code is unreadable at this point, too many flags and functions all over the place. I was forced to give up on it about a week ago and about 50$…

What I’ve noticed is that this is definitely not true AI — it has no real sense of context. It fixes one thing and breaks two more. For example, I had a transition with an opacity effect and a separate pixelation effect. The pixelation inherited the opacity transition, but the opacity had no pixelation. I spent an entire day trying to make the pixelation apply only to the pixelation effect while still combining correctly with opacity, and don’t tell me I did not prompt it right because I tried more than one hudred times, give it all the detrails in the works and when ask it do you undersntand it give the right answer only to mess up things with each iteration.

It would fix one issue and create two more. It doesn’t remember anything if what you’re building hasn’t been done before, which is exactly my case. It just hallucinates back and forth until you lose your mind. And of course, every time I say something is wrong, it replies, “Yes, you are right.”

Creating an app involves a huge amount of context and countless fine details that AI simply doesn’t understand. For it, everything is just a concatenation of words. The hallucinations get worse the longer you use it, and the larger the context becomes. In my case, by that point, I could not even take over as a developer anymore — there are red flags everywhere: functions inside functions, conditions triggering other conditions, and everything tangled into an unsalvageable mess.

So the conclusion is simple: never let it take full control. At that point, it’s game over — the code becomes unmodifiable and nearly impossible to fix, even for you as the developer.

I don’t see how this could ever be true AGI, since it clearly doesn’t understand context. Explaining an app to it is just feeding it concatenated words; it doesn’t grasp scope or how to fine-tune the small details that make a project truly good and finished.

But the propaganda works, all companies fall into this mess, thinking that now skill is not required anymore. If you let an agent loose inside a commercial or already working app or platform, it will destroy it — 100%, and if you ask for retribution, it will say,’ yes you are right :)’. CEO’s don’t understand this part, and of course, they want to replace everything so it’s just them and a bunch of agents that can “read their thoughts.”

So the conclusion, considering that this doesn’t really improve with new model iterations, is worrying. I could have used GPT-4.1 in my project and ended up with the same mess, not much difference. I expect the bubble to burst badly, because you can’t just pour billions in forever with no return yes, not much of a return if I, with 25 years of experience and solid skills towards creative development, failed so bad 99% of other devs, if not more, will fail the same or worse.

Yes, it can write an app from a prompt up to lets say 99.9% but what do you do with the 0.01% wihout that it will not work, and you will not be able to fix it as a developer due to the mess of code it writes. What is the point then?

This also reflects on all vidoes on YouTube with vibe coding, all tries are unfinished, look bad all kinds of bugs. If you think you can fix such a mess as a dev, well, you are wrong, my friend.

The only sane usage of this now is to use it for compressed tasks in the code and make sure you understand and follow it so that you can give strong guidance and do not let it go to lala land… and for that, you need solid skills, probably is best to do it yourself.

Is it good? I honestly can’t decide anymore. I feel as confused as the mess of words it produces.

Probably is a big fat lie at this point, I don’t see this getting better, the model by design is broken!

Another aspect of all their tests with new graphs showing improvements means nothing in the real world.

It’s funny that Sam Altman talks about curing cancer — really? More likely, AI could end up creating a new form of cancer that will be infectious and transmissible by air. This is a way better probability than fixing it. It feels like a disaster, destroying things around it and making the world worse overall, while the promise sounds like a big, fat lie.

I am so f*king angry of wasted a week on this project, but at least I understand now how this piece of s’it ‘AI’ works…

As for replacing me as a developer, that will never happen. The more I use it, the more I see what this really is — propaganda designed to attract billions more in funding, because apparently, it’s never enough.

3 Likes
2

You have valid points.

But let me play the devil’s advocate. I also went trough frustrating phases, and the blame was also on me. Because my lack of boldness, creativity, and understanding of this tool. So here is my tricks to improve.

AI are very sensitive tools, especially the conversational ones (most common).
The output is 100% dependent on the methodology. YOU NEED A PLAN! Use the AI for this to understand how it perceive a problem (the abstract)

  • Reset a conversation often to avoid self-poisoning.
  • Accurate coding terms can narrow the model’s field of view. You’re manager, AI is the coder.
  • Acurate concept/abstractions terms expand the model’s field of view. This is your main job as human, give the right amount of space but not too much
  • They are very good at fetching data. Do not hesitate to shamelessly paste links of any resources, even those using different coding languages.
  • These publicly available models aren’t made for research. They will not find new paths, only walk the known ones.
  • At some point, free mode isn’t gonna make it. The amount of processing is just too much.
  • However paying absurd money for AI is a scam. It doesn’t scale well for large task, only benefit is it doesn’t sleep like humans.

It’s a tool, it doesn’t replace skilled people.

2 Likes
3

In my case, it literally replicated the issue step by step. I spent an entire day asking, “Do you understand?” The answer was always yes, but every time it tried to implement the code, it went back and forth.

I’m not building a basic app. What I’m trying to do hasn’t been done before — it wasn’t trained on this. It’s a mess. You can say whatever you want about it being useful, but this is not AI — far from it. It feels like a fancy word completion system the more I use it.

4

Well, this is what I see all day long, left and right on YouTube. Also, my sales as a freelancer went to s*it. People are losing jobs like crazy, and believe me, most CEOs are convinced that this is truly AGI — it can do anything you can imagine. For example, “build me YouTube while I buy a new car.” Really, LOL…

I said it from day one: even with AGI, each app has its own context, with its own little things that can’t be prompted. Yes, it can help, but as I said, in my case it got me to 99% with my app and then got stuck. I tried to follow the code — holy s*it, it’s like spaghetti madness in there. It’s impossible to follow up at this point.

5

Eh, that’s the problem :wink: /s
If anyone could build innovative software using AI. It wouldn’t be a bubble under rentability floor.

6

You’re right about the mess AI creates when you let it drive.

But it’s not total garbage. There are legit use cases for it:

  • "What does this code do?” – good for parsing unfamiliar code

  • “Add JSDoc to these functions” – handles tedious work

  • “Write tests for this module” – decent starting point. Who likes writing tests anyway?

  • “Generate the HTML boilerplate” – saves typing

  • … and many others…

The “yes, you’re right” responses are infuriating until you learn to ignore them and/or laugh at it.

Your conclusion is correct: use it for isolated, trivial tasks where you can verify the output immediately. For anything complex, you’re wasting your time.

Most importantly, if you’re a beginner, stay away from it or you’ll become… a beginner with the delusion of competence. Not a good look :smiley:

5 Likes
7

I totally agreed

8

I feel bad for interfering so monolith discussion with my iconoclastic opinion. It all reminds me of the situation of a sculptor who blames the chisel that the sculpture got ugly. My point is, if you know the power and the limitation of a tool, when to use it and, more importantly, when not to use it, then you are the master of the tool, not its slave.

6 Likes
9

Agree, but you see that bilionaris CEO’s are advertising this like right now it write 90% of their code, well it can write 99.99999% of the code, even 100% if it does not work or worse, it works with hidden bugs that you are not even aware are there since you did not built it!

I don’t see how this technology will improve, to be honest since GPT 5 I saw no improvements; most of the time, it feels worse, but the way they advertise it, like in a few yers cure cancer, how if it is incapable of innovating anything, we are not able to kill a friking virus. They want to kill a cancer cell that is completely different in each patient!

The technology is not even close to what it’s advertised as, and it’s time for some of us to speak up; they have turned the world upside down with this s*it… I don’t see this working long-term; there is just too much money invested in this, and the result is mostly unusable in production code or anything else real-world. It can hallucinate and destroy you bussines in five minute and it will say after you ask it, ‘Yes, you are right’…

I am not against it, with all its issues, it helps me a lot, but I am against the way they are advertising it. This will replace this and that, and is not ready to replace nothing at this point!

1 Like
10

Maybe you can try posting this on the open ai discourse rather than three.js? https://community.openai.com/

1 Like
11

Personally - just me - I like the discussion here. I respect the opinions of people who frequent this place much more than many other places*. And it is - I think - an important discussion that affects many of us in this group.

edit: Just to clarify, there are so very many opinions about almost everything from almost everyone. And that is not bad, but I most interested in the thoughts of doers.

2 Likes
12

I completely agree. It’s extremely useful for analyzing a project’s structure and giving me ideas for how to implement new features. I usually don’t allow AI to edit code directly.
Honestly, I’m often surprised by its ability to come up with independent and creative solutions.
What matters is how we use the results the agent provides, not itself.

13

I am not saying that it’s not good. I’m saying that it’s being advertised as something that can replace me as a developer — or lately, almost all jobs — and that is a big, fat lie. Every time I let it take control of a project, it’s game over. The project becomes unfinishable, the code turns unreadable, and whatever fixes you add sit on top of other bugs, eventually turning into mega-bugs.

In one project that I failed to complete, it created 27 boolean flags for a bug it couldn’t solve — only one was necessary. These flags were scattered all over the codebase; at that point, it felt like solving a puzzle just to understand what was happening. And what we have right now is not AI — it’s a slot machine. Sometimes it works, sometimes it doesn’t.

It is extremely good for small, contextual tasks, but it does not understand the broader context. And none of these CEO idiots who promote it ever talk about that, of course.

It is impossible to vibe-code a creative, innovative application using full or partial vibe coding or leave AI take charge of things. If you don’t believe me, try it and send me the final result — I will find a lot of issues. It might work, but not as it should have been when done by an experienced developer. When it comes to creative work, it’s game over.

So my main issue is the way it is advertised. It is not what they claim, and I never heard one of this CEO’s speakign the truth about this part, they are manipulating the masses because they don’t understand how it works — it’s not even close. And even the latest models, like Codex 5.3, show the same problems. I don’t see this improving much at this point, because it doesn’t understand the actual problem or context.

14

Yes, that’s true. I think the disadvantages of AI agents will become more apparent over time, and most managers will recognize them but I also think they’ll keep using them anyway. :slight_smile: :slight_smile:

1 Like
15

which is why AI succeeds where you do not care for the details. do you think OpenClaw guy would care if the slider opacity mixes with pixelization correctly? if you want to micro-manage, use a human, not AI.

3 Likes
16

As a general advice, never trust ads. It doesn’t matter what they advertise, clips or cars.

I made a small experiment, and asked Grok. Here is its reply:
https://x.com/i/grok/share/4f091c613beb437f880a836ad29cf784

2 Likes
17

I have enough experience to understand what’s true and what isn’t when it comes to my job and life in general. After nearly 30 years of coding, building more than 50 applications, and working with over 50,000 clients, I’ve seen the industry evolve in countless ways. But the masses don’t have that perspective.

Everywhere I look, it’s “vibe code this” and “vibe code that,” claims that AGI will arrive by 2028, and endless hype from people who don’t even have a clue how to begin building such systems. It’s propaganda that’s damaging our industry, pushed so that soulless CEOs can squeeze out a few more billions.

In my case, sales of my apps have dropped by 90%. They’re almost dead since this nonsense started. None of my apps can be created with AI, and the same applies to all approved apps on platforms like Envato or the WordPress directory. Getting a WordPress plugin approved is more difficult than getting a job.

If you don’t believe me, give it a try: create a plugin, upload it to Envato, and then to WordPress.org. You’ll quickly see how high the standards really are.

Be aware that if your idea, execution, or code has issues, you can receive a hard reject. This means your plugin is permanently disapproved — you can’t upload it again; it’s lost forever. A few years ago, I spent 16 months working on a WordPress theme, only to receive a hard reject in two minutes, I can only tell you that I felt like the entire world was colappsing on me, oh and you will also not receive any info about the hard reject, just a copied and pasted text something like “Thank you submiting. bla bla bla, we appreciate the submission and hard work, generic stuff” so you have no clue why it was hard rejected, and yes I worked another two years, redesingning and rethinking everything… and got it approved (Acora), so almost four years to get a WordPress theme approved, is insane, I will lnever do that again for sure!

And yet, the propaganda works — “everybody can be a developer.” My grandpa, who doesn’t even know how to type, is supposedly going to create the next element in the periodic table while sleeping. Come on, let’s be serious here.

…

Junior developers are also being pushed out because of this mess. Most IT CEOs are not developers or engineers — they bought into the propaganda: you just type a prompt and see code magically appear.

The industry overall is falling apart, while AI company CEOs seem focused on replacing human work entirely. And of course, there’s very little we can do about it, I am trying to do my part I know that this tools are here to stay but we need to know the truth about them!

Skill will be necessary no matter how AI evolves and this is the part that all this AI CEO’s are avoiding like fire… heck they even said there is no point to learn much now, AI will do everything for you, you don’t even need a brain just give us everything!

With the project I mentioned, I’ll admit I got lazy. By the time I understood what was happening, it was too late — the codebase was in lala land and completely unsalvageable. But the experience gave me a clear understanding of how this tool works, I already knew this, the result was just the confirmation, even if vibe codding gets you to 99% that 1% (and no you can’t ix this 1% because you have zero clue how everything works, is AI code after all , a mess of code all over the place)… will never be done and that can be the hard reject that I told you about…

We will never have AGI with his models and this models are broken by default, they will always be that way simply because they do not understand the actual problem/context and this will not change unless they invent something else but I doubt that, saying that they will cure cancer made me lugh, we can’t even create a pill to kill a darn virus :)… but from all who heard them yes we will cure cancer in x years most believe it, this is the trap…

This mess is destroying our society, only so that rich, unethical people can become even richer.. no regulation are inplace is like wild wild west!

Are these tools useful? Yes, they are. But we need to speak the truth about them. Those who understand it — will it make any difference? Probably not, but my conscience will be OK.

18

Yeah, I can understand your frustration. I had the same when I realized that Santa is a big fat lie. And this does not stop companies to monetize on pretending Santa exists.

image
image720×451 61.2 KB

2 Likes
19

dark/white switch does not work, btw

1 Like
20

dont you use version control

2 Likes