I am using AI to add the funcitonality godot misses. to make tools in the godot to make it more usable. stop thinking you are a judge or something. even if it ai that asks, i see no problem in replying if you know the answer to the problem. We have ai as a tool to fix our godot project and take it further. your distrust in the ai tool is irrelevant
This thread isn’t about distrust of AI tools. This is about people (such as yourself), who clearly put so little effort into asking a question, that you didn’t even bother checking your post for half a day after you posted it, with the AI telling you what to post text still in there.
6 Likes
i even admitted it. i am not used to using forums at all. i was frustrated, i admited it lol.
2 Likes
I agree with @tibaverus the OP was about how to deal with the numerous examples of badly written AI posts with badly written AI code seeking help with little effort.
Given these are often as much troll questions as real (or so it appears) the OP did ask for thoughts on how to best answer them.
Most will ignore them and not respond. This is probably the best approach IMHO, but a polite suggestion and pointing to something in the docs might help too. Or a gentle nudge to do a tutorial or something first.
I mean, even when I do use AI, I take the time to express the question as best as I can. As was rightly pointed out on here (can’t recall who or where), using AI to help is fine, if you already know what you are doing. If you don’t, then AI will quite happily take you down a path that ends up in a complete mess of garbage code. Posting it here and saying “how do I fix this” is not really a good use of anyones time. Especially if the poster can’t be bothered to even explain what they were trying to achieve, or what they have tried. Here’s 300 lines of code, fix it!
And I bet every forum is suffering from similar AI slop bloat too. Not just here.
PS @flamenco687 did raise a good point though. Just because you start on AI learning, doesn’t mean you won’t end up in the right place eventually. I started when you had to go to the library to get a book. I can’t imagine doing that today. Things were slower then I suppose. Oh the good old days of just not having any answers to anything ever!
2 Likes
Fair enough. I admit I had bias in interpreting what you said in that way and I apologize for that.
1 Like
I do not have a problem with that. In fact, I think that’s a really good use of LLMs.
I believe you are projecting that thought onto me. I’m not trying to judge you for using an LLM in this post. You are probably unaware of this, but I knew when I linked to your post (and the others), that the link would show up at the bottom of your original post. I wasn’t trying to hide anything here.
Fair enough from your point of view. I think the answer to that personally is, I have no interest in making LLM models better through my time and effort solving their problems. I have better things to do with my time. However, I do like teaching, and helping people learn how to program. It brings me a lot of joy. I feel that fixing an AI’s code only helps the AI become more prolific, and does nothing to actually educate the person asking the question. Keep in mind, no human here owes you an answer. We are here as volunteers trying to help one another.
Having said that, I did, in fact, answer your question completely and fully in #6 of my answer. Not only did I tell you that the AI was giving you false information, I told you how to actually make it work. I can’t help it if you missed that.
Now I admit, I typically answer questions with step-by-step instructions (sometimes including screen shots), but in your case I gave you enough information to feed into your LLM, because I did not want to take a bunch of time doing its job.
But if you have questions about my answer, I’m happy to answer them.
We do. I’ve started using Claude.ai recently because it’s been in the news the past two weeks about how good it’s gotten and I wanted to test it out for myself. I find it quite helpful in certain situations. Like writing batch files for me. Or playing around with complex shader math. But I also know how to read the code it outputs, and decide whether it’s good or not. I also know how to test it’s output, find issues, and tell it how to fix it. I also know how to call it out when it makes up functions that do not exist. I have the knowledge to know when it’s wrong, and the vocabulary to correct it.
Again, I feel that you are projecting the idea that I distrust AI onto me. My post is not about people using it, or posting their questions. It’s about getting feedback on the best way to reply to them in such a way that is helpful, rather than dismissive.
If your post has started with, “AI told me to post this”, you would have gotten different responses, at least in tone if not in content. Sure, I was snarky, but I did still try to help you. I apologize for the snark. But 4.4 (Dev/Beta) is not a version. The version number is shown at the bottom of your editor.
And let’s be clear, your anger at AI for giving you made up information isn’t a reason IMO to have a crappy post or project your frustration onto other people. We are people with lives and games of our own to create. We do not owe you anything. Which is why the majority of responses in this thread so far has been “just ignore them”.
I appreciate that. LLMs are programmed to be nice to you and make you feel good. Us humans are human. Hopefully we can be helpful, but we do have feelings and baggage that LLMs do not.
4 Likes
Yeah, I wanted to clarify that, it’s just that I didn’t want to stray away from my message during the reply. As I’ve said, I sympathize with all of these doubts around AI. I’m naturally not against the people that have no clue/have valid points but to the fact the AI bubble is provoking this kind of discussions and activity.
As AI will never tell you and as humans usually say around forums: Search before asking, you’ll mostly find your answer.
Truly, it’s not out of rudeness but out of kindness for everyone’s time and benefit.
5 Likes
AI code should be the liability of whoever generated it, if they don’t understand the code they shouldn’t be using it.
5 Likes
This is so very true. When you let the LLM write the code then you are responsible for reading and understanding it. You cannot delegate both and develop any software.
5 Likes
I’m generally sticking to the ignore side, not only for vibe coded train wrecks, but for everything that looks low effort. Responding to such posts is a waste of time. Even if you do eventually untangle the mess, a member of the low effort brigade will soon get stuck again in the same way. After several iterations, chances are they’ll completely give up their little excursion into “game dev”. The time you’ve put into helping them will be for nothing.
The only reason for engaging may be if the problem itself tickles your hacker/sudoku braincells and there’s some opportunity to widen your knowledge or exercise your problem solving muscles. Unfortunately, most of the mis-vibed questions are from the desolate island of failed CharacterBodyXD boilerplates.
I often do give the benefit of the doubt though by initially responding with a short, probing one-liner. If the OP starts engaging and appears to be curious about things, I’ll give them more attention even if their code looks generated. I reserve writing actual code only for the people who seem to be genuinely applying themselves.
3 Likes
That’s a very good point. I’m actually quite taken aback by how complex they get so quickly.
Ehh,I am going with 2nd one.
Why you guys start a cold-forum war everytime with some AI stuff?
3 Likes
2 is fine as well but can be perceived as “confrontational”.
Yeah, it’s uncanny. A person “builds” 500 lines worth of code, with “everything” in it; vector math, collisions, inheritance, state machines, you name it… Then they suddenly grind to a complete halt trying to add some trivial piece of two-switch logic.
Strange times we’re living in 
3 Likes
It’s like They want results without effort.. . I also spend starting 5 to 6 months on ai to make my game .. but this was just disasters. Now i am learning evrything.. and it’s fun 
5 Likes
Man, I’m sorry you got misled like that.
I may not like or watch tutorials that much anymore, but they are a far more effective means to start learning.
I learned by using Blueprints (a visual programming language) in Unreal 4 and 5 to built a prototype for my old friends.
It was so hard for me to do at the time and I didn’t know much, but I naturally wanted to come back again and again, because I like the concept I was making so much. (It wasn’t Monkanics at the time, but it was a very similar 3rd-person shooter game)
(I think that’s the only legit response I gave to this topic, because I didn’t read the topic name after clicking a embed link)
2 Likes
I don’t get your center point.. it’s topic about ai .. and i don’t understand what you are saying sorry 
I’m replying to the top and deliberately ignoring the drama in the middle.
My take, having written games for far too long now is that I’m 100% on board with helping people who ask for help, if I know how.
If people are asking because they used an LLM and it screwed them over, I’m still going to try to steer them to the correct solution, because everyone starts somewhere.
If an LLM is asking the questions, I’m probably not going to help.
(footnote: I think I would actually cringe if I could see the questions I asked developers on cserve about 65k asm back in the day… )
4 Likes
Almost 50 replies already. 
2 Likes
#2 And make it site wide. Besides, nobody should be using (fake) “AI” to code until well after person has lots of programming experience under their belt. Then…. if they have that, they wouldn’t have any need to ask questions.
2 Likes
I’m assuning the majority of people responding here are involved in game development because of an interest in coding game mechanics, effects, and so forth. I’m from the other side of the aisle, with a long background in commercial art, including about thirty years of experience in 3D, motion graphics, and so forth. I also teach it at the university level.
In the course of my using 3D applications, I’ve picked up some scripting abilty, and through using Unreal Engine, I’ve gotten pretty comfortable with Blueprints as a way of creating and adjusting game logic. I’ve done a few online courses in Processing, ActionScript, JavaScript, C#, Python, and of course GDScript.
But coding has never been my primary interest. It’s simply a means to an end for me to create the images and experiences I want to create. I’ve found that Claude Code is really useful for me to rapidly iterate on my ideas in Godot. It’s tireless in answering my stupid questions and really good at explaining the “why” of every step of the process. It definitely produces a lot of crap, though, and the process of getting useful code out of it requires a lot of back and forth with the thing, proposing different approaches, debugging, and testing. I’m learing a lot about how to structure a project code-wise from it, although I always read the relevant documantation and make notes in order to verify its assertions. I would certainly never ask on here for someone to use their time to solve a problem created by me and my AI. I suppose I could work with a human coder to produce my projects, but I don’t have enough money to pay them enough to do good work that they have no interest in, and I have no interest in collaboration as such regarding my ideas. It’s different in an environment that is explicitly collaborative from the start. It’s similar to how I feel about coders who ask artists, “could you do all the art for this project I’ve got?”
I feel pretty much the same way about AI used for visual art as you seem to feel about AI used for code. Most of it is crap. It’s trained on everything: the good, the bad, and the ugly. And human production being what it is, the majoriy of stuff is bad and ugly, so that’s what you get out of it. Sometimes, that’s “good enough”. When it isn’t good enough, though, you need to be able to articulate why that is the case and know the solution, if only in general terms. With visual art, I can see why things suck, and don’t work, and I could explain it to people who are invested enough to listen and then learn why their AI’s production has failed them and how to repair it, either through manual intervention or sensible iteration through the AI. But they have to know enough about how visual art works, or be willing to pick up that knowlege practically and quickly for it even to be worth my time explaining it. Otherwise, they won’t understand why fixing that one thing merely exascerbates problems elsewhere in the project.
Many people will use asset kits, or prebuilt code to do their stuff, and that’s been generally accepted for years. Just look at the Godot asset libraries. AI is just another level to that. If you just want to slap together something to see what it can do, that’s fine. Admirable, even, since most people aren’t even that motivated/curious. But it isn’t enough to make me spend my valuable minutes correcting your work if that’s all you’re doing.
I propose that, as a community, we value our time as artists and coders and game designers, and be welcoming, as long as it isn’t a valueless waste of time for the people here. That devalues the human experience of learning and teaching. Ultimately, we’re interested in human development in learning to create interesting interactive experiences that will entertain and possibly educate others. We have to make human judgements about where and how to use our limited time on earth to help others, prioritzing those people who we think will actually benefit, even if we only say, "Get gud”. Choosing to not respond is sometimes the best thing.
3 Likes