Have any of these economists ever tried to scrape by as an entry-level graphic designer/illustrator?
Apparently not, since the sort of specific work which one used to find for this has all but vanished --- every AI-generated image one sees represents an instance where someone who might have contracted for an image did not (ditto for stock images, but that's a different conversation).
> every AI-generated image one sees represents an instance where someone who might have contracted for an image did not
This is not at all true. Some percentage of AI generated images might have become a contract, but that percentage is vanishingly small.
Most AI generated images you see out there are just shared casually between friends. Another sizable chunk are useless filler in a casual blog post and the author would otherwise have gone without, used public domain images, or illegally copied an image.
A very very small percentage of them are used in a specific subset of SEO posts whose authors actually might have cared enough to get a professional illustrator a few years ago but don't care enough to avoid AI artifacts today. That sliver probably represents most of the work that used to exist for a freelance illustrator, but it's a vanishingly small percentage of AI generated images.
There is more to entry-level illustrators than SEO posts. In my daily life I've witnessed a bakery, an aspiring writer of children's books, and two University departments go for self-made AI pictures instead of hiring an illustrator. Those jobs would have definitely gone to a local illustrator.
I've seen the same more times that I can count, having been in that business decades ago. Then it was clip art and bad illustrator work, no different than what you're seeing today with AI -- and to a trained professional, the delta between the two "home made" approaches and professional ones is clearly evident. We'll look at the AI slop in 10 years the way we look at clip art from 1995.
> That sliver probably represents most of the work that used to exist for a freelance illustrator, but it's a vanishingly small percentage of AI generated images.
I prefer to get my illegally copied images from only the most humanely trained LLM instead of illegally copying them myself like some neanderthal or, heaven forbid, asking a human to make something. Such a though is revolting; humans breathe so loud and sweat so much and are so icky. Hold on - my wife just texted me. "Hey chat gipity, what is my wife asking about now?" /s
I miss the old internet, when every article didn't have a goofy image at the top just for "optimization." With the exception of photography in reporting, it's all a waste of time and bandwidth.
Most if it wasn't bespoke assets created by humans but stock art picked by if lucky, a professional photo editor, but more often the author themselves.
I don't know. Even with these tools, I don't want to be doing this work.
I'd still hire an entry level graphic designer. I would just expect them to use these tools and 2x-5x their output. That's the only changing I'm sensing.
Probably not, economists generally stay in school straight to becoming professors or they’ll go into finance right after school.
That said I don’t think entry level illustration jobs can be around if software can do their job better than they do. Just like we don’t have a lot of calculators anymore, technological replacement is bound to occur in society, AI or not.
It looks like the writing is on the wall too for other menial and low-value creative jobs too - so basic music and videos - I fully expect that 90+% of video adverts will be entirely AI generated within the next year or two. see Google Veo - they have the tech already and they have YouTube already and they have the ad network already ...
Instead of uploading your video ad you already created, you'll just enter a description or two and the AI will auto-generate the video ads in thousands of iterations to target every demographic.
Google is going to run away with this with their ecosystem - OpenAI etc al can't compete with this sort of thing.
People will develop an eye for how AI-generated looks and that will make human creativity stand out even more. I'm expecting more creativity and less cookie-cutter content, I think AI generated content is actually the end of it.
>People will develop an eye for how AI-generated looks
People will think they have an eye for AI-generated content, and miss all the AI that doesn't register. If anything it would benefit the whole industry to keep some stuff looking "AI" so people build a false model of what "AI" looks like.
This is like the ChatGPT image gen of last year, which purposely put a distinct style on generated images (that shiny plasticy look). Then everyone had an "eye for AI" after seeing all those. But in the meantime, purpose made image generators without the injected prompts were creating indistinguishable images.
It is almost certain that every single person here has laid eyes on an image already, probably in an ad, that didn't set off any triggers.
Given that the goal of generative AI is to generate content that is virtually indistinguishable from expert creative people, I think it's one of these scenarios:
1. If the goal is achieved, which is highly unlikely, then we get very very close to AGI and all bets are off.
2. If the goal is not achieved and we stay in this uncanny valley territory (not at the bottom of it but not being able to climb out either), then eventually in a few years' time we should see a return to many fragmented almost indie-like platforms offering bespoke human-made content. The only way to hope to achieve the acceptable quality will be to favor it instead of scale as the content will have to be somehow verified by actual human beings.
> If the goal is achieved, which is highly unlikely, then we get very very close to AGI and all bets are off.
Question on two fronts:
1. Why do you think, considering the current rate of progress think it is very unlikely that LLM output becomes indistinguishable from expert creatives? Especially considering a lot of tells people claim to see are easily alleviated by prompting.
2. Why do you think a model whose output reaches that goal would rise in any way to what we’d consider AGI?
Personally, I feel the opposite. The output is likely to reach that level in the coming years, yet AGI is still far away from being reached once that has happened.
1. The progress is there but it's been slowing down yet the downsides have largely remained.
1.1. With the LLMs, while thanks to the larger context window (mostly achieved via hardware, not software), the models can keep track of the longer conversations better, the hallucinations are as bad as ever; I use them eagerly yet I haven't felt any significant improvements to the outputs in a long time. Anecdotally, a couple days ago I decided to try my luck and vibe-code a primitive messaging library and it led me in the wrong path even though I was challenging it along the way; it was so convincing that I wouldn't have noticed hadn't my colleague told me there was a better way. Granted, the colleague is extremely smart, but LLM should have told me what was the right approach because I was specifically questioning it.
1.2. The image generation has also barely improved. The biggest improvement during the past year has been with 4o, which can be largely attributed to move from diffusion to autoregression but it's far from perfect and still suffers from hallucinations even more than LLMs.
1.3. I don't think video models are even worth discussing because you just can't get a decent video if you can't get a decent still in the first place.
2. That's speculation, of course. Let me explain my thought process. A truly expert level AI should be able to avoid mistakes and create novel writings or research just by the human asking it to do it. In order to validate the research, it can also invent the experiments that need to be done by humans. But if it can do all this, then it could/should find the way to build a better AI, which after an iteration or two should lead to AGI. So, it's basically a genius that, upon human request, can break itself out of the confines.
People already know what the ads are and what is content, but yet the advertisers keep on paying for ads on videos so they must be working.
It feels to me that the SOTA video models today are pretty damn good already, let alone in another 12 months when SOTA will no doubt have moved on significantly.
Apparently not, since the sort of specific work which one used to find for this has all but vanished --- every AI-generated image one sees represents an instance where someone who might have contracted for an image did not (ditto for stock images, but that's a different conversation).