While I agree with this in general, I think there's a place for AI in conjunction with icon sets like these. Control nets give you the ability to manually design the meaningful details and AI fill the remainder.
As someone who works in the space and uses these tools, I've not seen any output without significant manual modification that was production quality, and I've only heard developers or non-designer AI-enthusiasts say otherwise. I know people who are about as close to AI->Prod pipeline as you can feasibly get right now in games, but there's still a whole lot of Adobe in that process, even for chintzy mobile games. Not to sound arrogant, but developers judging whether or not an icon system is sufficient for its intended purpose is like a designer judging whether or not a codebase is technically solid for a given use case. People who don't produce work like this at a professional level love these tools because they abstract away most of the finer-grained decisions that artists and designers must make when producing images. But without that granular control, even with all the tweaking and inpainting and whatever else you can muster up, that AI is still making more choices about what goes into that composition than the prompter, and mitigating that is often more hassle than it's worth. Text-based AI tools just don't have the consistency or the specificity required for commercial art, and maybe never will. There's just a really big conceptual disconnect.
I also work in this space and use it already a lot. Super time and cost saving. Just less need for graphical designers. A bit of after care sometimes needed. But you dont need to be a designer for that.
Care to share a workflow? No doubt I've seen people make shitty assets more quickly, but nothing I'd consider useful for any of my purposes. Besides, the hard part of design is figuring out the best way to solve design problems visually, not making the assets. You can definitely save money on designers if you don't care about having good design.
To be clear: I didn't say that you don't know anything about design— I said that I've never seen a direct generative AI to production workflow that yielded acceptable professional-level results, and any time I've seen people saying otherwise, it was because they don't have experience evaluating design. You countered with a specific claim and no details. I asked for the details: a workflow that would change my mind. Since you haven't yielded anything, I'm not really sure what you're looking for here. I can't prove I that haven't seen something.
Many designers could cargo cult their way through modifying a WordPress instance, but it would be patently ridiculous for them to assume they knew code well enough to have their unsupported assertion about software development stand on its own against the word of an experienced, educated engineer. There are many ways to accomplish the same thing with code, but not all of them are good, and their suitability isn't entirely subjective.
Most facets of design aren't subjective, either: there are technical components that you can't just intuit your way through, and there's more wrong answers than right ones. I am qualified to assess design because I am an experienced, professional, University-trained designer that uses these tools regularly. Not everybody is, and I'm not going to sit here and pretend any random lay person's unsupported opinion rivals that. Sorry if that bothers you, but I'm always open to being proven wrong if you've got anything concrete to add.
> You countered with a specific claim and no details.
I did not. You're responding to different people.
A lot of great indie game developers started out as designers, and I think the overlap of skills are much greater than you think. Likewise it feels like gatekeeping to claim that coders are intrinsically bad at design.
> I did not. You're responding to different people.
My mistake.
I know exactly how much overlap there is. I was a full-time developer— mostly back-end web developer but not exclusively— for 10 years before I took the 60 old credits I had going for a computer science degree and started my BFA. While I've done a little completely non-technical design, like branding and identity and print design, most of what I've done is technical, and still do straight dev work from time to time. I've had one foot in each world and am professionally experienced in both, but mostly work in game engines these days.
The first programming book I read cover-to-cover, back in the late 90s, was Learning Perl by Larry Wall. He famously said the three most important characteristics of great developers are laziness, impatience, and most importantly for my point, hubris. I see his point— if you're overly conservative about estimating your capabilities, you probably won't attempt genuinely difficult challenges. That hubris helped me push myself, grow as a developer, and solve some cool problems. It also tends to make developers unwittingly look utterly cocky when talking to subject matter experts. More often than not, developers conflate understanding how computers interface with a problem with understanding the problem space itself, and since there is so much overlap, that happens with design more than anything else.
I've got probably 10k hours under my belt contributing to FOSS, but unless it was a project I helped maintain, zero (0) percent of that is design work... Even though I could do the job from research, to planning, to implementation. I actually know quite a few hybrid designer/developers for whom that's also true. I've certainly tried, but I haven't in years. Why? Because pretty much every FOSS project is maintained by developers, and almost without exception, they think their interfaces and processes are great, if quirky and a little ugly... But it's no problem as long as users RTFM. In reality, they're just used to the shitty interfaces, and because they lack experience evaluating design, they think that means they're objectively good. That's why you see commits in many repost for user-facing FOSS tools saying things like "addressed UX criticism by implementing custom color themes" which is absolute nonsense. It's also why the only user-facing FOSS tools popular among non-developers are controlled by corporations or foundations with project managers and professional designers. Left to their own devices, developers make interfaces that are only tolerable to people with a working mental model of software, under the hood.
The fact is, most developers that assume they know something about design only really know how designs are implemented technically, and for the actual design part, they're just winging it, mostly unknowingly mimicking designs they've seen. They tend to assume that design is mostly a matter of aesthetics and have no idea what the design processes entail, what problems design can and can't solve, or the tools designers wield to solve them. They build to satisfy their own use cases and aesthetic preferences without knowing how to evaluate their fitness for real use.
Gatekeeping would be saying that you needed formal training to be a designer, or that you're not a designer unless you use certain tools or processes. I knew a self-taught designer that became the art director for a national magazine and much of his creative process involved sketching things out in hyper card on a decades-old Mac Classic he had in his apartment. However, making the assets that a designer would make without doing any of the work that makes it design is not design, just like cargo-culting 4 deep nested for loops with database calls in a WordPress template is not engineering, even if the end result doesn't seem much different to lay people.
