I don’t think anybody said that. I mean there is plenty of stuff automated without AI so such a claim would be rather weird.
I was quoting you according faster and AI audience and also:
AI to the rescue..
..why ?? It was done for thousend of years with HI (human intelligence)..
..of course.. there are some applications for AI.. but i doubt that it is the only way to do it.. or even needed
in somethign like blender as the OP sugested on RCS ..so:
- Can Blender 3D have an AI ?
- Yes.. it’s open source.. do what you like..
.. but the blender devs may be busy doing something else.. and maybe also don’t want to have some trouble like some artistic social media with AI.. so they just try to improve blender itself.. 
It is difficult for me to understand what exactly you have in mind. Could you explain your points with actual examples, describing what you have and what you would expect the AI to do?
Well, yes. But I am sure that for some things AI is a better method than other methods. So, imo, if AI is suited better than use that. If other methods are suited better then use these methods.
1 Like
That’s what i’m talking about… for what is actually AI needed here ? It’s just a general suggestion…
Just for the sake of using it ???
( just recognized this could be out of context because of the thread merge… )
Just replace Blender 3D by any other app… ( or detergent )
… so what ??
And there are already usage of ai for blender…
( …i also wonder why there wasn’t any other reply from the OP… so it is even a general and not blender related waste of time ?! )
Well, it is my understanding that an AI is better than other methods or at least has the potential to bet better than other methods for turning a wide range of generic or imprecise requests into specific things.
I guess that is what the poster wants. Kind of a digital assistant that does a wide range of stuff the user.
Hmm… well, I guess it is kind of a Captain Obvious suggestion which every user and developer is aware of so in that regard suggesting it on RCS might be a bit redundant.
Exactly the opposite. AI is trained on specific things and even if trained on multiple: you have to be precise so that it could generate something maybe usefull… or have to refine your requests… and then check if it works… especially: if it works in general or just for the actual example…
So wouldn’t it be just simpler to :
- learn some of the knowledge mankind has already mined in thousend of years… and
- use it with this incredible neural network we are all born with… which
- already comes for free… and doesn’t need thousends of Kilo Watt
instead of:
- making oneself even more stupid/ uncreative and
- let the machines do the thinking
Maybe offtopic… but for the context:
( I always wonder how some people think their government want to mindcontrol them but then do the most stupid things… and even publish them on the internet… there seems to be no mind to control anymore – or: if i see some of the questions asked by some people on the internet… no help possible )
Back to blender and intelligence: some people are not even capable to have a look into the BlenderDocu which is accessable via Help → Manual…
…
So how could an BlenderAI help…
<sarcasm>
This was an rhetoric question…
</sarcasm>
Hmm… I don’t see how this is correct.
If you write and addon without AI you have to integrate every single thing you want it to be able to do.
If you let an AI learn how to use Blender and how to understand language then you can tell the AI to make stuff for you in Blender.
So how is an AI not general?
I don’t have to be precise either. I can just let the AI create variants and then select the one I like best or tell it to refine feature X. Kind of like what art directors do by instructing their human artists.
I don’t think it would be simpler. Even now I can create images with AI that I could never paint myself. Punching a couple of prompts into Midjourney is way simpler than learning how to paint.
Letting the machines do the thinking is a lot easier. If it is better is for everyone to decide. However, I am sure that when such technology exists it will be used because it makes the art creation process cheaper.
… but that’s not by some random anything… it has to be based on something… here…
you yourself said
Thats’ called refinement and you have to do it.
So my point regarding the original thread is (and you alredy wrote it yourself):
Not the AI is giving the solution but you have to do something…
This suggestion is to general… suggestion for possibility to use ai to make blender… to solve user task…
The kind of machine learning which I consider useful in an application like Blender are: Denoising of rendered images, image super resolution, frame interpolation, image to pbr material, generate seamless materials, generate seamless material for a specific 3d object, possibly style transfer. Further, motion capture like human images/animations to 3d poses (including face), motion tracking and many more I am forgetting.
In my view, most of those would help people to be more creative, because it becomes easier to create the assets, or it is faster to see how the result looks like. It opens the door for more experimentation.
4 Likes
That’s a precise description of some possibilties for using AI… 
I newer was argueing pro/contra…
Well, I wouldn’t expect an AI to read my mind and that I would not have to do anything at all. I don’t see how that is relevant, though. Just because a tool does not do everything for me automagically does not mean the tool is not useful.
I am kind of confused how you could interpret my point like that.
My point is that in the possible use cases I described, the thinking is done by the machine and yet, I don’t think people would become more stupid or uncreative because of it.
I’m not speaking of unusefullness of AI. I was only (and only that) talking about the unspecific suggestion of:
When you said (sorry for re-re-citing):
I think (my opinion) the point of AI is not to solve unspecific assignments. And you also give an example of an very specific task to do.. which i think does not need AI because it was done already before by artists..
( and i also was not talking about mindreading )
My basic question for this suggestion was:
→ Why and for what exactly do someone need AI for this (very vague) suggestion ?? 
And additionally i shared some toughts about: Why do someone think that users are not inteligent enough to learn simple tasks and this feature only can be done with additional inteligence? [..and how the heck humankind did this before AI 
]
And then i extrapolated/exaggerated ( 
) abit about users getting dumb to not think for themself..
So i’m talking just ( and only) about the senselessness of this suggestion.. ( and by far more then the OP)
And also think @anon98372585’s examples are totally valid (and they of course wouldn’t make people dumb.. in fact they are very smart use cases of AI.. unlike: to provide a user interface to solve user tasks.. ← 
)
I’m not fighting here.. just wondering about the assumptions in this specific suggestion by the OP about the need of AI to make.. 
.. well.. what exactly..
..and now: night is coming..
Ok, so but what is a user task? Everything is a user task until it is not, isn’t it? Back in the days making GI and AO were user tasks by faking it with specific light setups. Now we just press a button because renderers are smart and performant enough to do it for us.
So, imo, anything can treated like this. Instead of doing task X i just tell my digital assistant to do it for me. I can’t see many tasks that a human could possible achieve faster than a (future) AI unless they are extremely simple tasks.
The example with the ear I gave is not specific imo. It would need to work for “select all vertices of the nose” as well. And for the knee. And for the exhaust gas recirculation of the car the character is sitting in. And besides selecting stuff it should of course also be able to make a walk cycle, a drive animation and animate the camera in style of “insert your favorite director”. That is what I mean by “unspecific”. Ideally the AI should be flexible like a human but fast like a computer.
That imo is the point of AI in the long run.
Most likely is to help for batch-editing-tasks where typically now you need to write a script. Instead of having the AI generate a Python script, most likely is to get info-window commands directly.
This is what Adobe is currently trying to do, as well as Microsoft in Excel etc…
TLDR; but I saw the bit about “what is human creativity”, look what I found 
:
Many befenits if a plugin was created by some AI :
-
It is free , anyone can ask AI to create the same addon or a variation
-
Many specific tools and addons that are needed per project could be created with AI and most artists will never learn python or never code addons.
Other areas AI would be very usefull :
- AI unwrapping , able to creare great UVs using shapes recogniton, better results than auto retoplogy tools
- AI creating Geometry nodes , who likes to spend days creating spaghetty nodes ?
- AI creating great detailled base characters 3D models without user to have to study characters.
And so many other ares ( animation for example) where AI would do the heavy base work and would smooth a lot the 3D workflow.
I am impatient to find new ways in which the computer can help me to avoid drudgery. Therefore: If “AI algorithms” can help me do that, “I’m a fan.”
But at the same time, I know to downplay “the ‘I’ of this equation.” Because this is not, and is never going to be, “intelligence” in any human sense.
Instead, we are simply referring to “internally-directed algorithms,” in which the algorithm itself is used to optimize its own parameters within a suitably-constrained solution space. (The “constraints” having necessarily been externally determined … at least for now.)
I’m eager to have them “relieve me of part of my load,” but I know that they will never replace me. A power screwdriver can never tell me why the screw should be put “there and not somewhere else.” Nor would I ever ask it to.
2 Likes
The more concerning bit I have, is that the whole AI hype means the entire concept of writing an algorithm or even a simple for loop will become a lost art. Why would anyone actually write code or even open a code editor when AI can replace it all (even though AI will always be changing and will put you at the mercy of companies thousands of miles away and their TOS, as opposed to a traditional function where, absent of API changes, you know it will continue to work as it did many years from now)?
It is the same with ditching the idea of creating content with general creative software in favor of the generators, the people at Midjourney are likely the same as any other tech. company, in which they reserve the right to make any change to what is considered acceptable use for any reason they see fit. The type of work you do with it today may suddenly be declared problematic a few years from now.