Do you think AI could be used by TLG to design minifigures and even sets?

No. AI isn't logical enough to understand LEGO, and even if it was, you'd spend more time writing the prompt than would take you to design the same thing. It's a total waste of time, money, and energy to even try at this point.

In short: LEGO sets designers jobs are totally safe for the foreseeable future.

Edited June 27, 2024Jun 27 by Murdoch17

To add to Murdoch's points, AI can't account for the qualitative factors of a set like playability and the build experience. It could probably make some decent looking scale models, but there's no way it can account for the factors that make Lego special.

Now, promotional art or advertisements, I could see them using it for that. AI is a useful tool but it's not a shortcut to creativity. 

AI can be used by TLG to help design minifigures and even sets.

I'm a writer in the marketing and communications office at a university. No, we do not use AI to do our work for us. Yes, we do something use AI to help give us some ideas. It's a tool, just like a calculator or CAD or word processing.

Incidentally, there was a builder at some of the larger shows here in the US last year and the year before, who used AI running on a jailbroken Mindstorms device to write haikus by tapping a Lego brick onto a Mac keyboard. (He is @legosaretotallyforadults on Instagram.)

17 hours ago, kerryball said:

Do you think AI could be used by TLG to design minifigures and even sets?

Minifigures to some degree perhaps yes if we're talking color combinations and prints, sets definitely no. The details would require an endlessly long explanation, but suffice it to say that all current A.I. models just aren't there yet. This is going to take years before we even come close to what an experienced designer can do.

Mylenium

I do not believe so and I'd be very reluctant to buy something not designed by humans.

2 hours ago, Mylenium said:

This is going to take years before we even come close to what an experienced designer can do.

That: "Years".

And to further add: Can do "within the limitations of a profit oriented company". We are not talking about "freely experienced designers can do" approaches, are we?

Furthermore: What are "years" in a lifespan of a human? Or a company's life?

I have no doubts, >none<, that AI will change the "LEGO" world. Maybe in "years". Well, I believe it is actually happening right now. Maybe not in the TLG bubble, who knows, which was established (but apparently did not much change since then) more than 60 years ago.

And then: What is wrong about such change? The fallout (as in "what do we 'get' from the gods") of new LEGO sets is certainly controlled by the anticipated extent of commercial success, isn't it? I mean, there is competition out there. When the application of AI leads to raking in more money, do we actually believe, TLG won't do it???

Well, just my 2 cents, essentially worth nothing!

Best,
Thorsten

 

5 hours ago, evank said:

AI can be used by TLG to help design minifigures and even sets.

I'm a writer in the marketing and communications office at a university. No, we do not use AI to do our work for us. Yes, we do something use AI to help give us some ideas. It's a tool, just like a calculator or CAD or word processing.

Incidentally, there was a builder at some of the larger shows here in the US last year and the year before, who used AI running on a jailbroken Mindstorms device to write haikus by tapping a Lego brick onto a Mac keyboard. (He is @legosaretotallyforadults on Instagram.)

I disagree with this completely.

An important part of the brainstorming/idea forming stage of design is being able to mindfully select inspiration and know the provenance of any concepts. This is important for any artist but especially for a corporation like Lego, which will want to avoid accidental plagiarism or copyright/trademark infringement. And frankly, generative AI as it currently exists is useless for that purpose. Pretty much all major generative AI platforms are trained on databases that include plagiarized material, and none can effectively trace the results they generate back to their original sources.

Frankly, the best ways to come up with concepts and ideas are still good old fashioned research and development. AI is a lazy and irresponsible means of skipping that step, at best useful for amateurs without the skill or means to put forth the appropriate effort but of no use for skilled designers and concept artists or for a company with the means to hire the appropriate human talent.

I wonder if the OP is a bot:/

6 hours ago, Toastie said:

That: "Years".

Well, so you are a click worker who's telling some A.I. we haven't yet heard about how those hundreds of elements go together with each other in infinite combinations? Congratulations! You're my hero! Snarky comments aside, people just seem to completely not understand how current A.I. models work and just jump on the hype train. Yes, it's here to stay and we can argue over whether it's going to be years, decades or just a few months, but for now I'm firmly convinced that it's still a ways off. This is a whole different discipline than generating an image based on a few billion scanned other images and applying a style transfer algorithm or other such "simple" use cases.

Mylenium

3 hours ago, Lyichir said:

An important part of the brainstorming/idea forming stage of design is being able to mindfully select inspiration and know the provenance of any concepts. This is important for any artist but especially for a corporation like Lego, which will want to avoid accidental plagiarism or copyright/trademark infringement. And frankly, generative AI as it currently exists is useless for that purpose. Pretty much all major generative AI platforms are trained on databases that include plagiarized material, and none can effectively trace the results they generate back to their original sources.

Frankly, the best ways to come up with concepts and ideas are still good old fashioned research and development. AI is a lazy and irresponsible means of skipping that step, at best useful for amateurs without the skill or means to put forth the appropriate effort but of no use for skilled designers and concept artists or for a company with the means to hire the appropriate human talent.

Allow me to disagree with your disagreement. Yes, there are serious legal and ethical issues, but where ideation is concerned (that's what we folks in the creative business call it), A.I. can be a source of inspiration. An A.I. doesn't care whether it's been trained in "good taste" or academic rules of whatever trade and that can be an advantage. It's like your mind going off in a dream and coming up with the wildest ideas. I've seen stuff generated by A.I. that I never would have thought of myself and as much as I may not like stealing other people's content, at least to that end A.I. can be a useful tool. I may re-create the image using my own methodology and refine it based on my artsy background, but it doesn't change the fact that it was inspired by something that I saw elsewhere. And that is kind of the point: At the end of the day, why should it matter whether I draw inspiration from a "real" photo, something I see in my own environment or a synthetically generated image? I feel you cannot dismiss A.I. in that regard. Whether in the end it's worth it to sift through thousands of generated images to find the right one or start from scratch with old-fashioned R & D, as you put it, is an entirely different question.

Mylenium

2 hours ago, Mylenium said:

people just seem to completely not understand how current A.I. models work and just jump on the hype train

Maybe some others do. We use it now and then in our research projects, works quite well from time to time. Certainly not always - does mostly make no sense at all using it - but there are sometimes such complex datasets, we experimentally recorded, that we simply do not have the slightest chance to "classically research" the meaning behind it. And yes, you may ask: "Why do you do such weird things in the first place? Make it simple, stupid." Good question, good advice, my answer would be: "We tried for more than 10 years narrowing down the experiments to lift the complexity - and failed."

Does all this have anything to do with "creativity", "imagination", "talent", "research and development"? Surely for the latter - for the former I don't know, as I don't know how they "work".

Best,
Thorsten 

I reckon AI could be used to design LEGO sets, if trained on existing builds and a 'goodness' score could be defined could be defined to used to filter the acceptable designs and use these to iteratively train the model. The real problem here is defining what makes a good model to use as the score to automatically train the model without human input. I could see it going two ways, one in which the operator tells it what to design - a house, a car, etc - and the other to let it design something totally freely - where you might get a lot of wacky ideas such as sailing ships with wings, houses on legs, etc. Although probably not so wacky given Dreamzzz exists.

For minifigures, it could probably be used now. Some years ago, people were using AI to produce minifigure images. But given how conservative minifigure print designs tend to be, I doubt they really need AI to generate artwork for the prints. 

If AI is used at TLG, great, there are lazy designers. 

But why “Legoland” in the title here?

26 minutes ago, Lion King said:

But why “Legoland” in the title here?

Its not in the title. It says LEGO *and*. No second L.

I don't think AI is useful for designing with Lego.

First off, with minifigures, yes I can see where AI could be used but it comes back down to 'why bother?' The iconic minifigure design is something that has been consistent for 40+ years.  The main body remains the same, just with different printing and different accessories. The real design innovations are the ideas or themes of what to make the figures (and even then, if they are licensed, the follow the design of the IP being used). We don't need AI to give us ideas like, 'lets make a goat farmer', or 'lets make a e-gamer girl.' The recent space CMF series is a good example of this, the designers said they came up with 100 minifigure ideas for the series and the real work they had to do was to reduce that number down to the 12 they eventually released.

Now for sets, I do understand and acknowledge that the arguments get more complicated. First off Lego designers have been successfully designing Lego sets for decades without the need of AI help. And also, i won't go into the issues already discussed already like play value or the building experience.  The biggest current problem with AI is that it will generate something that looks good, but not realistic.  Great care needs to be taken for AI to not use bricks that don't exist, or placing the bricks together in impossible ways. Or designs that would be too flimsy or unstable and would fall apart due to gravity if made in the real world. If you factor in all of these checks and edits and human adaptation of what an AI could come up with, then you end up spending more time and more money on a design that a Lego designer could've done in the first place.

Sorry, I hope this rant makes sense, been having trouble finding the right words as I'm writing it.

 

Also, AI wouldn't be able to hide some of the Easter Eggs that Lego loves putting into their designs.

2 hours ago, Murdoch17 said:

Its not in the title. It says LEGO *and*. No second L.

Ah ok. Thanks. My vision is really bad at the moment.. i misread that title.

On 6/27/2024 at 9:48 PM, Lyichir said:

at best useful for amateurs without the skill or means to put forth the appropriate effort but of no use for skilled designers and concept artists or for a company with the means to hire the appropriate human talent.

That may be your opinion but it's not at all how smart organizations in creative fields are operating right now. I live this every day. My team consists of a half-dozen talented, veteran writers. AI tools have become an important arrow in our quiver. They are * just * tools, like spell checkers or Photoshop or calculators. AI won't take your job -- someone who knows how to use AI will take your job.

I would bet a large sum that AI is also being tested, if not already part of the daily workflow, among designers at Lego headquarters. They'd be foolish otherwise. Chances are, technical staff there wrote or are currently planning their own GPT, and it's likely trained on their own database of designs.

 

On 6/28/2024 at 11:29 AM, MAB said:

I reckon AI could be used to design LEGO sets, if trained on existing builds 

But how many actual models are out there, including LEGO's competitors and MOCs? Arguably an infinite number, but at the same time not enough. A considerable part won't even be available digitally. And even if you assume there would be enough digital models to train an A.I., you'd have to vet them beforehand or else the old "Garbage in, garbage out." bites you in the butt. Unless someoen already has been working on this for the last three years or so I don't expect any results soon.

Mylenium

On 6/28/2024 at 2:43 PM, DelQuinn said:

The biggest current problem with AI is that it will generate something that looks good, but not realistic.

Not really a point with a brick design, don't you think? How real is real? Even the best models are just simplified approximations.

On 6/28/2024 at 2:43 PM, DelQuinn said:

Great care needs to be taken for AI to not use bricks that don't exist, or placing the bricks together in impossible ways.

Not too big a concern. The process would be limited to a pre-defined pool of pieces and you could even restrict the number of uses per element time. That could be part of the algorithm or a secondary process based on its own A.I..

On 6/28/2024 at 2:43 PM, DelQuinn said:

Or designs that would be too flimsy or unstable and would fall apart due to gravity if made in the real world.

Again, not really an issue. Engineers do it all the time already for "real" objects and machines in their CAD programs. Same as above - it could be an extra step or algorithm that communicates with the main A.I. and provides optimization data or ditches models that don't meet certain requirements.

On 6/28/2024 at 2:43 PM, DelQuinn said:

If you factor in all of these checks and edits and human adaptation of what an AI could come up with, then you end up spending more time and more money on a design that a Lego designer could've done in the first place.

We can agree on that one. Current A.I. wastes a lot of time and resources for unusable results and for a complex system like a LEGO model that would be exorbitant. You could let an A.I. run for days and have it spewing out model after model and none of it could be anywhere close to what you expected.

Mylenium

Edited July 1, 2024Jul 1 by Mylenium
Fixed typos

Well ...

On 6/28/2024 at 2:43 PM, DelQuinn said:

First off Lego designers have been successfully designing Lego sets for decades without the need of AI help.

True, but times change ;)

On 6/28/2024 at 2:43 PM, DelQuinn said:

AI to not use bricks that don't exist

It won't because it should not be allowed to do so ...

On 6/28/2024 at 2:43 PM, DelQuinn said:

placing the bricks together in impossible ways

It won't because it should not be allowed to do so. A lot of this stuff is already known to Stud.io.

On 6/28/2024 at 2:43 PM, DelQuinn said:

Or designs that would be too flimsy or unstable and would fall apart due to gravity if made in the real world. If you factor in all of these checks and edits and human adaptation of what an AI could come up with, then you end up spending more time and more money on a design that a Lego designer could've done in the first place.

I carefully disagree here: Of course there will be tons of these designs. But I believe a LEGO designer will see that right away. The AI works essentially for free. Experienced LEGO designers are really expensive. The usual stuff, I know. But ...

Now, why do people believe an AI algorithm, a program, will totally design the set? Box, art, instructions, no flimsiness, all legal connections, the whole bang? Not the point. Such an algorithm may come up with really cool "hulls" - well, "may", duh - just go to reddit or any other current so cool outlet ... Yes, there are soo many pictures but hey, maybe the entire - so educated human assessment process should kick in. "AI" is generally a computer program, and (at best), should be currently viewed as a tool. Back in the days, Word 5.5 form MS was called totally evil. As was Win1.0. Things apparently evolve. There seem to be some survivors and diehards still using MS Word on MS Windows platforms.

12 hours ago, Mylenium said:

Current A.I. wastes a lot of time and resources for unusable results and for a complex system like a LEGO model that would be exorbitant. You could let an A.I. run for days and have it spewing out model after model and none of it could be anywhere close to what you expected.

Yeah, true, of course. But how about having experienced humans looking at all the trash? Define exorbitant - maybe your AI should not crank out trash all the time. Maybe a little office space with experienced designers should do the training. I really don't want to how many folks in the far East may do that ... and then imagine, the trash rate goes down. Programs only spew out model after model, when told to do that.

No, I am sorry, LEGO bricks are real, their numbers of combinations are >absolutely< clear. They seem to be endless, but so is a PetaByte. And a trained AI is not a random piece of software, going into the wild. Training is one thing. Assessment of the results is so much more important. And yeah, algorithms can do that as well. Crunching through an existing, ultimately complex LEGO build is a joke for a suitably equipped AI machine. What it does, or we do, with the "analysis" (yielding another crazy dataset), and even what the analysis tells - I have no clue. But hell, this is what Google does every nanosecond ...

It is all flowing as of now. That is my take. For me, the key thing is how we plan the interaction between "AI" and human's - well, let's call it - ingenuity.

Best,
Thorsten   

17 hours ago, evank said:

That may be your opinion but it's not at all how smart organizations in creative fields are operating right now. I live this every day. My team consists of a half-dozen talented, veteran writers. AI tools have become an important arrow in our quiver. They are * just * tools, like spell checkers or Photoshop or calculators. AI won't take your job -- someone who knows how to use AI will take your job.

I would bet a large sum that AI is also being tested, if not already part of the daily workflow, among designers at Lego headquarters. They'd be foolish otherwise. Chances are, technical staff there wrote or are currently planning their own GPT, and it's likely trained on their own database of designs.

 

Your argument relies on the assumption that organizations getting into AI are doing it for smart, informed reasons, rather than because it is an overhyped product that makes investors excited. You have failed to describe a use case that is more than just a lazy shortcut for idea generation with inferior results to actual human-driven concept work.

Tools like spell checkers or photoshop or calculators are things that have definite use cases—assisting development in ways that humans without those tools could not do as efficiently or effectively. Unlike something like a calculator or spell checker, which are used to intelligently root out human error, or something like Photoshop, which simplifies editing tasks by granting editors more deliberate control over their image files, removing the human element from design exploration undermines the entire endeavor by outright eliminating the deliberate human choices of the creative process. The point of design exploration is to consciously explore possibilities and push them toward a finalized state through iteration, and existing GPTs are fundamentally incapable of both of those tasks (lacking conscious understanding of the subjects it was trained on, and incapable of iteration due to not being able to recall and learn from its past outputs). For the purposes you are proposing it for, it is nothing more than a creative dead end.

Smart and informed reasons: it saves a lot of time, and sometimes it gives us ideas that we hadn't otherwise considered (for example, there was recently news in computer science that AI figured out a better sort algorithm, after decades of people thinking that we humans already knew the best ways and that we were at a roadblock).

Investors? I work for a public, non-profit university.

It sounds like you have long made up your mind with an arrogant opinion that humans, or at least you, are perfect and that technology can't help.

You will be very disappointed when you inevitably hear someone at Lego say in the news, "We love AI, it helps us generate awesome new ideas!"

As I said, I'm a professional writer. I have about 25 years of experience. AI is not a "creative dead end". It's opening all kinds of new horizons for my team, and more teams at smart, informed organizations around the world. I would fully expect Lego to be one of them.

21 minutes ago, evank said:

Smart and informed reasons: it saves a lot of time, and sometimes it gives us ideas that we hadn't otherwise considered (for example, there was recently news in computer science that AI figured out a better sort algorithm, after decades of people thinking that we humans already knew the best ways and that we were at a roadblock).

Investors? I work for a public, non-profit university.

It sounds like you have long made up your mind with an arrogant opinion that humans, or at least you, are perfect and that technology can't help.

You will be very disappointed when you inevitably hear someone at Lego say in the news, "We love AI, it helps us generate awesome new ideas!"

As I said, I'm a professional writer. I have about 25 years of experience. AI is not a "creative dead end". It's opening all kinds of new horizons for my team, and more teams at smart, informed organizations around the world. I would fully expect Lego to be one of them.

I don't think humans are perfect and that technology can't help, I certainly don't think so about myself. Please don't put words into my mouth.

But there's a world of difference between the mathematical or engineering problem-solving of creating a sort algorithm and the largely creative, aesthetic skills required for something like Lego set design. And that is where AI does not (and arguably CAN not) exceed human creativity, because ultimately all it is attempting to do is IMITATE human creativity, and (so far) generally doing it remarkably poorly. Expecting an AI to come up with an artistic design, even in the concept stages, is a misaimed pursuit because a neural network is incapable of knowing what works aesthetically and what doesn't, let alone why certain things work and certain things don't.

And ultimately, it's trying to solve a problem that doesn't exist. Human concept artists already exist and for the most part, those with actual skill and talent do not want or need generative AI to automate away the work that they both specialize in and genuinely enjoy (especially given how unethical all major art-based GPTs are when it comes to plagiarizing training data). If I do hear Lego designers praising AI for concept art generation, I will indeed be disappointed, not because I was wrong about its use case or lack thereof, but because it would represent a cheap, dubiously-ethical shortcut for something Lego has proven perfectly capable of doing with human talent for decades upon decades.

By saying "those with actual talent do not want or need generative AI," you're insulting me and my team. And by saying that anything created with the help of AI is "a cheap, dubiously-ethical shortcut," you're just being incredibly closed-minded. Word processors didn't destroy writing, CAD didn't destroy drafting, and Photoshop didn't destroy art.