Ask HN: What am I supposed to do after I’m “disrupted”? Work in video and CG
Now I am seeing the video generating AI (imagen video from Google on the front page) and I’m 100% sure a good percentage of the work will vanish. There’s loads of work that ad agencies will just hire an AI prompt guy to generate for ads. Big companies will still make ads, of course, but smaller gigs that keep the whole industry afloat? What about even 10 million $ shots with CG characters that will now become commonplace?
I’ve retrained before. I started off as an editor, then did VR video, now moved into CG. I’m pretty good at my job - I’ve worked on stuff for Dell and Apple, including stuff you’ve probably seen.
It’s funny to think I might have to retrain yet again. I didn’t expect image generation AI to be the next big leap. I was already moving into more storytelling content (ie documentaries) because that’s more defensible against AI. But I expected 3-4 years before video generation would just come out.
Now it seems like it’s happening so quickly I’m not sure they won’t have good stuff out in 6 months to a year.
So - HN - what do I do now? And what about the other fellas in my industry who will be out of a job? In my estimate, we are talking about tens of thousands.
To counter the obvious: I have learnt web dev at one point I thought I’d make the jump, even started learning React but I found the work to be mind numbing. I just love making images. But I feel like Pierre Auguste Renoir’s who was a plate painter in France and did a good living out of it - until the process was industrialised and he struggled (maybe for the rest of his life? I don’t quite recall).
To counter the second suggestion: AI prompt design is not image making as far as it would interest me. I also don’t think there will be a huge learning curve. Becoming a filmmaker on my standard requires about 10 years experience or an excellent school (of which there are maybe 10-20 in the world). I would imagine AÍ promoting will be done by the lowest paid interns available.
What do I do HN?
224 comments
[ 2.9 ms ] story [ 241 ms ] threadPersonally I think there's a good chance you're overestimating the the extent to which your day-to-day job will actually change by incorporating AI. Machine learning was supposed to take over data science years ago, but data scientists are in more demand than ever, and their work often doesn't involve too much actual machine learning.
Also I asked the question a bit as a challenge yes I do have my own ideas on what I’ll do.
I sympathize with you a lot here.
But it's not really a new problem that great art usually doesn't have commercial value commensurate to effort or quality.
It’s bizarre to me. I’m 33 years old and I still used 16mm and 35mm film at film school. When I was there people were still saying “digital cameras will never be the same”.
Then RED cameras came out and the world slowly moved to digital filmmaking. Then I saw things moving towards CGI and what it once took Pixar 10 hours to render I can do it in better quality on my laptop in 10 minutes these days. I enjoyed that, I thought that was awesome. There’s still skills involved. You have to light the scene even if it’s virtual. Virtual cameras work the same way. I take pictures on medium format film for practice.
Now it’s like - everything I’ve ever done in my life can be done by a ten year old and a sentence. It’s … frightening, awe inspiring. Many things at once. Almost overwhelming. It makes me think we’re at the end of an epoch, because it seems to me that we’re about to automate almost everything.
Realistically, making images by hunching over a computer for 10 hours is not a profession many can retire in anyway. Like coding (or, perhaps even more so, because the pressure to deliver is actually muc harder), it's a young man's game. People burn out mentally or even physically - carpal tunnel, back pain, neck pain or just lower energy levels due to age can prevent you from putting in the required hours. Realistically, if you stay in the industry, in 10-15 years you'll move into management or high-level consulting anyway.
"We use AI to optimise our early-stage iterations"
"AI is great for generating initial concepts. Here are some of ourfavourites, and some that didn't turn out so well!"
"Did you know AI can reduce our turnaround for clients by up to 50%?"
In other words: it'll help. It won't replace you.
Society just needs to accept a whole of our social truism is politically correct LARP to the drum beat of a financiers profit memes. What are a bunch of rich elders going to do if billions quit dancing like jesters for coin? Fight us from their far off estates?
800k sworn LEO in the US working for a paycheck, not fealty to a billionaire. The public has all the advantage but isn’t allowed to know it. Communal health is the future of work whether outsider financiers accept or not. We’ll just flood the inter tubes with AI media and crash the value of their copyright constraint system if they don’t like it. I have zero obligation to do-nothing daddy deep pockets who can point to a contract I wasn’t a party to. I don’t believe such people are worth coddling with my agency.
I’m using AI to aid development of a game engine with a baked in model to replicate and help the user evolve the game world style, and rules, to their curiosity and tastes. Games will be even easier since familiar experiences are heavily constrained through painstaking play test. Game world generation parameters are finite. Even the art work we appreciate most is often constrained to a handful of style patterns; consider how flat web design was suddenly everywhere with only colors differing. Whole lot of the same goes into game asset production.
Gabe Newell called content creating AI an extinction level event for multimedia business. Of course politics as usual will be leveraged to create private monopoly of such technology unless open sourced.
Our best bet is such software is open source and thus aristocrats cannot lock it up behind artificial scarcity and copyright maximalism.
It’s funny to me that even here in HN people don’t get this at all. They fall behind simple truisms “automation only increases jobs”. It’s almost as if they’ve forgotten Trump got elected because vast swathes of America were deindustrialised and that led to a collapse to the quality of life in those regions.
On research that I’ve read, automation has definitely decreased the number of industrial jobs in America. Those people, some in large numbers, never found other jobs that were as well paying.
But like David Simon always says “American distopia doesn’t sell in America”.
Whether they're worth taking Vs doing something more similar to your current job but for a bit less money is an individual's choice, of course.
New histories of the period think of it as potentially as one of the few times we know where there was an enormous economic boom alongside a general collapse of quality of life.
Ir could be we are looking at something similar at the moment with the digital revolution.
Yes eventually we all retrain and the effect will be nil, but in the in-between time, a lot of people might suffer immensely.
We're talking about things that can be eased into, by white collar workers. And the numbers affected are nothing like the industrial revolution's either.
Driving seems like a tough nut to crack, but what if we suddenly start obliterating multiple areas of human employment one after the other? Ones we haven’t even thought of?
Which would you pick? Keep in mind you don't have to go through the headache of actually hiring people - all you need to do is not fire them.
Effective prompt design is a very deep subject, and one that is changing all the time.
My optimistic version of all of this is that people like yourself will be able to continue doing the work you are doing today... but better, and faster. You'll be able to produce higher quality work because the generative AIs will be able to speed up a whole bunch of the more time-consuming tasks for you, allowing you to spend more time on the creative and editorial side of things.
I believe people with genuine skill, talent and taste will produce better work with the assistance of generative AI than people that do not have that existing background.
I might be wrong! I hope I'm right.
It's a nice experiment, but I really doubt the level of artistic direction required to meet specific customer requirements will ever be replaced by "AI"
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.
For this I think you should try to be one of the 10% that's left. You're on HN and you're really early in this AI image generation game.
One example: I'm running an early-stage startup. If an explainer video for my product will cost me $20,000 I'm probably not going to commission one. If it costs $2,000 then maybe I will.
I've seen plenty of talented, experienced artists on various AI subreddits who are incorporating prompt-driven images as a small part of their overall work.
A relevant example: Newgrounds banned AI art: https://waxy.org/2022/09/online-art-communities-begin-bannin... - but they included this exception:
> There are cases where some use of AI is ok, for example if you are primarily showcasing your character art but use an AI-generated background.
It is not easy to do "prompt engineering". Keep in mind that most generated material looks awful -- there is a huge selection bias going on.
The outputted stuff needs to made production ready also, distribution, etc.
If anything, the emergence of these technologies should greatly expand. You will have a new customer base that was not available before, and you will be able to deliver new products.
Another example is the uncanny valley of CGI people: at what point was it acceptable to audiences, and at what point was it excellent?
It's pretty early to declare that a victory for human drivers isn't it? I'd be worried, if I were them.
You spelled "two years" wrong.
Yeah. Look at machine translation - it's gotten pretty good, in some limited cases it's perfect, but it's also frequently just clueless, and for anything serious you absolutely need a real human translator.
I think these AI tools will improve, but I'm extremely skeptical that it's going to improve by like an order of magnitude any time soon. It will still require extensive tuning, heavy selection, and accepting that you probably can almost never get it to make exactly what you want.
Perhaps this will just make your job easier and not impact your ability to make money. Maybe you will make even more money.
And if at some point the industry dries up completely, you will move on and do the best you can. We all will.
So at the very least the parent probably has time to do a bit of soul searching to decide and prepare for their next undertaking.
However, I don't think it'll replace much or even a lot of work, especially where a minimum level of quality is desired.
This is not without precedent. DesignJoy is run by a single designer who makes over $1.5 million a year by himself, doing contracts for 20+ clients at a time [0].
[0] https://www.indiehackers.com/post/broke-the-1-5m-arr-mark-as...
Reductio ad absurdium.
Are you as full of loathing for all the HN people saying “just integrate the AI into your practice” as I am? I like drawing stuff and I love that this pays my bills and I really have zero interest in becoming a “prompt engineer” instead.
I think a strategy of creating an image of AI art as cheap and tacky is useful. For how long, I don’t know. If we’re lucky then it’ll turn out that getting these things out of the domain of creepy claw hands is a lot harder than anyone thinks it will be, and there will be a lot of obvious tells for a long time.
It may also be useful to try and get your professional associations to bring some suits against these things for playing fast and loose with “fair use”. The legality of these things is debatable, and so’s where “fair use” should fall - personally I feel like nobody posting work on the internet anticipated their stuff being scraped and fed into a giant neural net designed to take their job. Make this shit a lot pricier by demanding that it be built on art explicitly licensed for machine training.
Figuring out where the intersection of “art you like to make” and “art the AI sucks at” lies is worthwhile. If your passion is realistic painterly work then you’re fucked, what do you enjoy doing that isn’t that?
I am not a lawyer but I don't really see the "it's not fair use" argument.
People have been inspired to write about subject matter because of their experiences.
AI takes lots of things that currently exist and mixNmatch.
It's not nearly the same thing.
why not? Just because people take in experiences slowly, via senses, doesn't mean that an AI isn't replicating that learning via a fast method.
If you answer no, then you must necessarily admit it's different.
It can do a lot of things, but in the end it's still just a glorified function.
What's the difference between retelling someone's story orally, and using a printing press to make an exact replica?
What's the difference between writing down a conversation from memory, and recording that conversation?
What's the difference between making a nude painting of someone, and taking a nude photograph?
What's the difference between going out in public among human beings, and going out in public where there's a facial recognition camera on every corner feeding everyone's movements into a centralized database?
What's the difference between the grandma who knows everyone in the village, vs the social media company that knows everyone in the world?
Social customs evolved in a context of fundamental, sharp limits to human cognition and skill. When technology smashes through those limits, those social customs don't work anymore. Copyright law didn't exist in a world where you couldn't copy things mechanically. If you apply the old rules naively, you end up with a nasty world that nobody wants. So we have to invent new rules to limit how people use the new technology, otherwise people get exploited and life becomes intolerable.
"What does copyright law say about this" doesn't even make sense as a question. Copyright was invented before AI was. The question should be, "what would a society someone would want to live in, say about this?"
But that's not what's happening. You can't "crack open" a model to find the bitmap of a piece of training data. It's not there any more than a painting you've seen is "in" your brain. A model sometimes creates things that look similar to existing pieces if it got a ton of copies of the same image, for the same reason that most people's artistic rendition of a tree is going to be more accurate than their rendition of an anteater.
The entire AI art debate is a symptom of the fact that society is structured such that labor-saving technological advancements can harm more people than they help, and it's baffling to see otherwise intelligent and ideologically similar people fixating on this single, relatively minor technology rather than the much more consequential, broader issue of the fact that automation always moves wealth into the hands of a smaller and smaller group of people.
In other words, a technology that has never existed before and doesn't follow any of the categories and schemas we dreamt up to organize the world. So we need to make new rules.
This is very much not a fact.
The average height of Englishmen actually went down during the industrial rev. presumably due to malnutrition. New histories of the period think of it as potentially as one of the few times we know where there was an enormous economic boom alongside a general collapse of quality of life.
Is it ok to scan copyrighted works. I think Google Books' win at the Supreme Court shows that yes it is.
Is it ok to process them down to the rawest information?
Is it ok for people to generate content from the raw information (and ok for them to charge for it)?
The last two I don't have an answer for, but I understand the fear of artists and the anger that their hard work is being shovelled into the monster that is going affect their work. But also more broadly the make-up industry, the lighting industry, studio spaces, lens makers, paint manufacturers, etc etc etc.
a compiler is not transformative. You translating a book into another language doesn't make it a new work.
But an AI that takes billions of images, and uses it to synthesize something different and new, is fairly transformative under my eyes, and deserves new copyright. Unless the AI generated image is largely composed of a small number of works, i don't see why anyone should have copyright ownership of such an output!
if anything I would argue that human taking code from multiple places and making it compile into something more useful is much more synthetiz-ing than AI...
Fortunately I enjoy writing code.
I still remember splicing together There Will Be Blood and enjoying looking at some of the beatiful still frames.
My boss at the time spliced together Quentin Tarantino's Grindhouse, and somehow fucked up and spliced an entire reel out of order... and nobody ever noticed (because Grindhouse was full of intentional fake projectionist mistakes, being an homage to an old era). He ended up noticing on his own on the second day of showings and fixing it overnight!
I make 100k a year in London making CG. I don’t see how that will be available to me doing AÍ prompt work.
[0] https://news.ycombinator.com/item?id=33099182#33100703
nor does the art that humans create - the emotion comes from the human viewing/experiencing a work. The "source" that work is perhaps irrelevant, if the human viewing it is being evoked.
You’re crazy if you think many artists I know don’t outwork 99% of the population. Condescending and inappropriate. Many artists I know are the most hard working people I’ve ever met - because making it in this area is fucking hard. And a lot of the laziest people I know are well paid data scientists. I don’t blame either of them. But your comment is just a lazy ridiculous preposterous idea. Read the biography of Patti Smith and tell me she didn’t work hard. Such a presumptuous comment.
And if you think the vast majority of the human population in history, subsistence farmers, didn't work harder than virtually every artist in existence, I'm not sure what to tell you. Being an artist for a living is absolutely a privileged position for 99.999% of the world's population.
BTW I'm sorry you might be getting disrupted. It truly sucks.
I think this is what people have in mind when they talk about integration. It doesn't seem ridiculous to me.
[1] https://weirdwonderfulai.art/general/stable-diffusion-integr...
Moving somewhere less expensive than London and doing your current work remotely might be an option, everything’s become less focused on going into the studio to work after the ‘rona. Lower your expenses and start playing with personal work, your client history can probably get you some interesting gallery opportunities to give you a decent chance at selling Authentically Human-Made work to rich folks at an absurd price.
I ditched your kind of career path a while back, I’m using the drawing skills learnt chasing the Hollywood animation dream to draw indy comics and furry porn. Makes less money but I can eke out a living in a much cheaper town than Los Angeles, and I pretty much draw what I want.
I'm not sure how to say this in a non-dickish way, but most of us have to do stuff we don't particularly enjoy in order to make a living. I get that it sucks from your vantage point, but like... I'm sure there were stablehands who loved horses and had no interest in becoming mechanics. Shit happens.
I'm a writer and equally in danger of automated replacement, but so it goes. Maybe I'll end up curating and editing GPT7 output instead of drafting everything myself — I can live with that. It's not like my paid work was written for fun in the first place.
The art I make for my own satisfaction, I expect to continue crafting word by word, and I think people will continue to appreciate that painstaking expression of the human spirit. Just not in a commercial context.
https://www.aei.org/economics/what-atms-bank-tellers-rise-ro...
This doesn't always happen mind you. But it shows that your (and my) economic instincts are piss poor. We can't just assume what the future holds.
-------
Another few examples:
3D animation has killed a lot of tweening artists (the people who smoothed out animations for Disney Films like 101 Dalmations or Sleeping Beauty). But 3d animation created the need of modelers, texture artists, riggers, and more.
Automatic Drum machines didn't kill drummers either, but allowed for more music to be made in general.
The only thing that automation "killed" recently was Lawyers, as online webpages that auto-generated common forms removed a ton of jobs that Lawyers used to do. IIRC, Lawyers are in somewhat of a decline because of this.
So it just goes to show that no one really can predict these things.
together with the previous comment on mechanics and stablehands, i wonder how many people in history grew up training and wanting to be a mechanic, only to be told, sorry we just don't have that many automobiles in the world, we need more stablehands. i wonder how many people went to do fine art in school only to be told 'look you have good skills and all but we just need more tweening artists'.
I think retraining to an adjacent field is definitely possible - the only thing i see disagreements in is the cost, and who bears that cost.
Whenever people try to support this argument, they either fall into the “not thinking quantitatively” fallacy by pointing out that “sure, 25 writers got laid off, but the company used to have no programmers, and now they have 2!” Or they fall into the “not explaining the mechanism” fallacy by giving examples from the past where “stablehands retrained as mechanics” without examining why that happened, and what aspects make 20th century mechanization and 21st century automation completely different in kind.
are you aware of how taxing that work is on the body ? How isolating it is ?
so you're asking those people who farm (using huge machines and automation to do so extremely efficiently) to give you free food? Why would they want to do that?
So, what aspects are completely different in your opinion? I don't see any fundamental differences. People's wants are unlimited, so the workers no longer necessary due to automation can move to fields where automation can't be applied. There will always be plenty of those because, again, unlimited wants. I mean, in Japan, people are already paying for artificial friends, who will go out for dinner with them. I think we'll have much more work in human contact and companionship in the future. Hopefully, much more doctors per patients and much more teachers per student than today. Etc.
Software jobs don’t scale like manufacturing jobs. If you automate a particular profession’s work, they’ll lose their jobs, but meanwhile maybe now 10x more customers can afford the service they provide. But now, scaling the automating software to serve 10x the customers won’t require 10x as many programmers, maybe just a couple good devops people to tend the flock.
You might then argue that customers will still find things to want from the newly freed up labor, but I also think we’re going to start being hard-pressed for categories of work where “the human touch” makes a difference, and also where there is room to employ large sectors of the workforce. Add in that customers “new demands” won’t necessarily correlate with the need for human workers, and it becomes easier to see a world where people keep coming up with new wants, and automation keeps swooping in and immediately satisfying the demand.
It's hard to retrain. When the change happens between generations there's a natural progression with parent: stable hand, child: mechanic.
When it happens 1-2-5 times in a person's career, then those are some really painful resets.
Sure, some will swim. But the social contract with democracy isn't that "some" swim, it's that "almost everyone swims" (which we generally call "everybody", but that's a separate discussion).
Long term and across society automation is absolutely positive sum. But short-to-long term for those on the short end of the stick it's clearly not.
it's not zero sum, because the resources that used to be spent paying you now could be spent on something else. This increase in efficiency means more goods/services could be produced!
Of course, that newly saved money would be spent on someone else, instead of the person being made redundant by automation. It is thus a societal responsibility to retrain/reskill that person, and the training perhaps also partially be paid for by the entity benefiting directly from the automation.
Don’t forget about musicians and junior software engineer work…very soon.
If you are drawing completely new images then I'm not sure how competitive these tools are. They're just serving a mash up of existing works. Sure they can produce something interesting, sometimes stunning, but usually not much better than what you could just find already.
As for producing a physical painting, that has its own constraints of material and ability.
For now I think it'll only replace some low level commercial stuff and perhaps add images where there were none before.
For each technological advancement we had in the past 200 years had made it easier to do things at a certain level. Which enabled people to do more things and more complex things. Which just pushed up the average level of what we create higher.
Can't AI be used to handle what the artists had to spent more effort to create before, therefore enabling artists to create more complex works of art and therefore push the creative level up?
Are you as full of loathing for all the HN people saying “just integrate the AI into your practice” as I am? I like drawing stuff and I love that this pays my bills and I really have zero interest in becoming a “prompt engineer” instead.
I wish it were that simple. As a person who genuinely enjoys being a data scientist, and practices all those skills in a project I create on my own: the industry says otherwise. Integrating a technology into your practice that takes nuance out of it, just really means that managers and c-suites are going to expect you to give more nuance to your choices (which they deadlocked you out of). ML DevOps for the last two years are just me integrating more services.
In the worst case corporations will just rebuild the A.I. models from data licensed from stock photo sites, getty, etc. All you can get from this is a one-time bulk-discount payment, and then you will be paying them for use of the model in the next Photoshop.
Plus there are copyright-abusing corporations that jump on any excuse to kill off fair use and scraping, so any attempt to legislate scraping may instead end up creating more Disney copyright and link-tax laws.
I understand your livelihood is being threatened, and that’s very scary. Can we please avoid using the legal system to kill nascent competitors? Surely there are better responses than this.
Burn it all down.
Finally people will get culture back. We won't have to appeal to corporate giants to have access to movies, images and music.
Him and his crew spend a lot of time together. They have deep discussions about philosophy, art and ethics while spreading smelly goop around concrete floors. I bet they go home pretty intellectually satisfied from doing epoxy floors.
Develop business relationships.
Good clients are mostly insensitive to price.
That’s what makes them good clients.
Good luck
You’re right that really high end work will continue, most likely unchanged, for at least a decade (maybe). But I’m not so sure really - things are moving so fast.
My point is on the quality of clients, not the quality of the work product.
In the future, clients who prioritize cheap will go on fivrr and get 500 AI alternatives to choose from if they can.
Or down the block to a lower priced competitor later this afternoon if they can.
Good clients want you to stay in business because they want to work with you again.
Is AI a tool or a trap?
Can it make some of your jobs iterate more quickly and weed out the cruft or heavy lifting?
Or is it only something that gets in your way of production?
It is up to you.
Because of this, there will never be a 100% reliable "text prompt to AI image" generator. Someone will have to get to know the AI and figure out how to make it do what is wanted, or modify it to get what is wanted.
Kinda the same way Google and other information repositories have not made researcher's jobs obsolete.
Just focus your efforts on continuing to produce quality work, those clients never dry up.
As someone who used to work in web dev, they most certainly do and did.
And you have an ever growing pool (worldwide) of workers who are jostling over that dwindling supply of gigs.
Many places that I've worked with are now considering using a zero code website creator instead of actually coding, and for many scenarios it just works perfectly, cheap, and flawlessly. Of course it's not for everything especially if there's customization and server technologies involved, but it's very big to create disruption.
Sounds rhetorical, but I ask sincerely.
I think junior software engineers could realistically be obsolete in 5 years. Imagine a software architect being able to prompt: "write a rest API that exposes user data from system A to internal system B", and have a complete API dumped out, with tests, ready to be deployed.
I don't believe this will scale to all software anytime soon, but the type of thing that junior engineers work on is vulnerable.
As a related thought, if not an upside, did you read that story about the guy who outsourced his job to China? He worked for some telecom and just hired out whatever they wanted him to do to randos overseas and basically ended up doing nothing other then going to meetings that set up scope.
I believe he was fired but - end result aside - it's not hard to see how this might fit into your situation.
Embracing the change means learning the new tools and adapting your job around them. Fighting against it means lobbying the government to create protectionist laws for your industry to prevent new technology from disrupting you.
The former is easier if you are more flexible and open minded. The latter is easier if you have a lot of money and are well connected. Neither is inherently easy, which is why most of us just complain instead.
maybe you could consider joining the machine on a project that feeds the AI with unique new art, I imagine the models will need to be updated or re-learned with "license friendly" art at some point, might be a nice way to go out with the bang..
I have yet to see a stable diffusion output that is high quality work. They often lack any compositional sense, and all the lines are weirdly wavy. It's like someone trying to do realism but with the motor skills and spatial awareness of a toddler. The picture that won the state fair awards looked really terrible on further inspection. Basic composition elements like the horizon line or the giant circle in the middle were all skewed, and not in an attractive way. Any elements beyond the central figures are just noise. I see that story as more about bad judging than anything else.
Stable diffusion has its own version of uncanny valley, and it may be a long time before we can cross it. The people who decree stable diffusion as the death of human art don't often to have a firm grasp on artistic or design fundamentals.
Depends on the widget. There are lots of categories of widgets where you just want to show happy people. Or if you're advertising a brand rather than a specific product, you may want to show happy people with a logo, but the products don't need to exist.
AI+human will produce better results than just AI for a long time but you have to learn how to use it.
One of the interesting possibility will be to train your own AI on your private creative work/dataset so others could not have it if they don't have access to your private dataset, and it would produce your style better than some general AI. There will be services that will do it for you as for example Emad from StabilityAI already mentioned they will provide something like that.
But yeah, if AI keeps improving the end game is a little scary.