I'd be happy to bet that the tech will continue to improve. But "producing a specific image with generative AI is sometimes almost impossible" seems very likely to still be true in February 2024.
A lot of commenters here are making a very common mistake with observing any burgeoning technology: assuming that its current weaknesses will be present indefinitely.
Assuming that current models' problems with, say, details and specificity, will continue on is equivalent to looking at 19th century automobiles and scoffing that they'll ever largely replace horses. Or looking at "smartphones" from the 90's and determining that they'll never see much uptake from the average consumer.
AI image generation models are going to continue getting better and better and better, and the flaws we see now will get less common and less severe.
I think it's still up to debate whether this tech is really a winner takes all sort of thing, I think even with the image generation ai craze, we saw a bunch of different implementations come out in succession. Maybe one might be slightly better but it's still possible to get something close pretty quickly
Look at what AI generated images were like a year or two ago. They were mostly abstract, dream like images that were cool but not usable. They have improved 1000x in a year. Bottom tier animation from cartoon images now, could be seriously impressive in a few short years.
As an artist who uses generation as part of my art (the rest is me) I find these images extremely cool, though I prefer non objective abstractions. Given the technology is still immature, I wonder if improving the technology will lead to less "art" — what we see currently are really imperfections in the process rather than deliberate choices. Maybe in a few years the end result will be more realistic and less fantastic. Will that be better?
In the future researchers and big corps will have much more heavily annotated data sets that can provide models with things like aesthetics and accuracy scores, and low quality AI generations won't impact output much unless prompted for (e.g. "early 2020s style AI art image with deformed hands and phantom limbs")
The tech will get better, but ultimately there still has to be a human who decides 'that's the one that looks good', which strongly depends on someone's taste and skill in identifying what a good image looks like.
There will probably be less need for designers of 'lower quality' simple images though.
I think it's too soon to see real data on this because the current leader in AI imagery is still quite limited in what it can produce. There are too many examples of stock photos where no combination of tokens is going to render a similar picture, there are still very prominent and noticeable limits on what AI can produce.
When creatives are seeking out a stock photo, they're not looking for a piece of generic clip-art to fill a void on a page. They're looking to satisfy far more prescriptive set of requirements, the sort of thing that should really be shot anew, but budgetary and timing constraints typically mean that a pre-shot photo will have to do.
As AI improves it'll make a bigger impact, even right now in its limited form, it's not looking good for junk-stock peddlers, which will disappoint absolutely no one.
We think that image-gen models are lagging behind LLMs by about a year, so the problems that these models will have should look quite different in the future.
It'll require us to be adaptable and also to take chances on solving problems that aren't huge issues just yet, but are likely to be once models improve.
I agree with you, that 3/5 is stretching. This seems premature.
But, at the rate we're seeing progress, I don't think there's any doubt at this point that top of the line models will be able to do all the proposed examples by June 2025. In fact, by June 2025 I bet that millions of people will be able to generate those images on their home computers.
AI imagery is here to stay and will get better every day.
A service should either embrace it or they will lose a significant portion of their users/customers to another who does support AI content. While most of the AI content is not ready for prime time in terms of coherence and resolution, it's just a matter of time that it reaches (and quickly surpasses) traditional methods.
With Stable Diffusion and similar generative systems we have seen a leap in generative art/media, partially with significant improvements within a few months. What makes you think this was the last or only leap in the next 5 to 10 years? As if progress would just stop here? Huh?!
Do you think we hit a ceiling were progress is only tangential? A line which is impossible to cross? Otherwise I dont get this mindset in the face of these modern generative AI systems popping up left and right.
It still takes a surprising amount of labour and artistry to get an AI to give you exactly what you want (see also: Pareto principle). Consumer expectations scale proportionally to technological progress, so demand for premium assets and brand differentiators won’t be going away anytime soon; the state of the art has advanced but stock photo business will advance right along with it.
Equipment may pose constraints on what you can feasibly do (though fewer than people think).
How and what you then do within these constraints is largely up to you.
Will AI increasingly get better and better framing, colours, processing, etc? Most definitely. For the immediate future though, the emotion and ideas of talented, innovative photographers will still stand above.
The crazy thing about generative AI models is they provide a service that's significantly better in both price/output and bests possible output.
In addition to being able to create completely novel, high quality images, they can do so at speeds that no human could ever hope to match. And image models of this quality have only been around for a year, imagine where they'll be in five.
I empathize will all the artists who feel cheated out of these models training on all their data, but the sad reality is these AI models are just far too useful to ever go away. The world's standards for art and text have gone up faster than they ever have in world history over the course of the last 10 months
Considering the staggering speed that image generation is improving, that 10% gap will only continue to close.
Starting e.g. an art education right now seems likely to be extremely nerve-wracking as your talents may very well be woefully obsolete by the time you graduate; the exception perhaps being those top-0.1% talents that will feed the models of the future with new material.
I think it probably will within a decade. Just to use still images, the tools that have come out with stable diffusion already allow a lot in terms of generating variations, inpainting, fixing faces, etc. Give these tools time to mature, models keep getting better and bigger, and hardware keeps getting better - you are absolutely going to replace still images "soon" and if you can do still images, video won't be far behind.
The guy who needs an image will write a prompt and paste it in to some tool. The prompt goes to a language model that's been fine tuned on those websites that share prompts and image-gen creations. The language model spits out nine variations of the original prompt that it thinks will improve the output. The nine generations plus the original prompt produce ten variations from the next gen, or next next gen, diffusers. The original guy put in his prompt and gets back a grid of 100 examples. Does he like any? Maybe mark a few, refine the prompt, mark a few more, get variations of the ones he's marked. Expand one or two, edit something out, add something in, generate another thousand variations, and he's got something really good.
If this process gets fast I think you'll see a few minutes from a non-expert can produce better illustrations than professional artists. I don't think this means that everyone will be a professional-artist-equivalent - just like anyone could deliver a pizza but not everyone is a pizza delivery driver. What it will mean is that getting professional artist output will become something that anyone could reasonably pick up and do if they do a small amount of learning to get the hang of the tools. Plus, just like you might deliver a pizza to your friends or family, if you needed to you could produce high quality art.
reply