Predictions about the impact of AI-generated imagery

Installing one mediocre camera as a wide-shot is a lackluster solution. I would never choose only one as a decision-maker in that role, but the people in those roles usually don't understand camera operation, the talent of a truly great sports camera operator, and/or even the flow of sports. So the idea of a set-it and forget-it solution is too fantastic.

And it will be...once the chips are always inside the balls and the cameras are better. But right now some of these high school sports streams look like security cameras following the action.
 
https://twitter.com/moons_artwork/st...kF8y8IJuRjHSwA

OP: If you are one of these three professionals, you need to understand AI tools today or be out of a job in 3 years: - Writer - Artist - Coder.
Learn to Prompt.

Art director: Art director here. I will never hire anyone who creates prompts.

I know and am friends with world class, best of the best concept artists who say they will never use them.
I won't ever use them, I don't need em. I get by fine.

Anyway, forget all the bullshit, learn your fundamentals, hone your craft

3rd guy: Are you being as hardline as you appear, or do I misunderstand? As an AD myself, I don't mind if my artists play around with AI art, if only to understand what the tech is and isn't.

Art director: Used as a tool, part of a workflow, no problem with it. Using that only and then calling yourself an artist and expect to be paid an decent wage? Never. Would laugh them out the door.

There's a lot of Dunning Kruger effect going on, where the people making the most breathless predictions about AI know nothing about the industries in question

There also seems to be a growing consensus that middling work (in illustration, writing, etc) is most at risk, but that quality work won't be done by AI. But for me, as I imagine it is for many, a decent amount of my bread and butter work is not that demanding (though it varies)
 
There's a lot of Dunning Kruger effect going on, where the people making the most breathless predictions about AI know nothing about the industries in question

Yesterday, I was looking up some info on Milton Friedman of the early 70's. Found an interview with three different Nobel Prize winning economists - him included - with predictions about the GDP growth for that year (I think it was 1972). As one might expect, there were three different answers. To quote from "Adventures in the Screen Trade", "Nobody knows anything...... Not one person in the entire motion picture field knows for a certainty what's going to work. Every time out it's a guess and, if you're lucky, an educated one".
 
To me, it doesn't matter either way. AI could take over everything right now or never be seen or heard from again.

But if I have to think about it I try to think about the bigger picture.

And with the above examples (and others), I cannot for the life of me visualize humans in a future world paying other humans to write, draw, paint, code, build websites, etc. under normal everyday circumstances. People will definitely still be doing these activities because they won't just stop overnight, but not accepting that technology will further be involved in our lives and careers seems foolish to me.

Maybe I'm so wrong and everything will be exactly the same as it is today on December 4th, 2022 - but I still cannot picture a world which functions the same way if you have AI capable of doing these tasks pretty decently.

___

After seeing some of the AI art/stock a few of you linked here or somewhere else (IDR), I'm sorry to say I would never hire a human artist again unless it was something personal and heartfelt, or AI could not get it "good enough".
 
.

And with the above examples (and others), I cannot for the life of me visualize humans in a future world paying other humans to write, draw, paint, code, build websites, etc. under normal everyday circumstances. People will definitely still be doing these activities because they won't just stop overnight, but not accepting that technology will further be involved in our lives and careers seems foolish to me.

Maybe I'm so wrong and everything will be exactly the same as it is today on December 4th, 2022 - but I still cannot picture a world which functions the same way if you have AI capable of doing these tasks pretty decently.

Have you interacted with chatgpt? It's very informative about how AI works, where it excels and where its limitations are

The thing about ai-generated art is that I think it's fine if you're good to go right out of the box. But if you need to craft it into something more specific, you'll need to combine additional skillets

Someone on Twitter posted some very interesting images from the prompt "vintage horror children's books". I'm not sure what AI they used. I tried the same thing on Dalle2 and my results weren't as interesting. But I liked the illustration below a lot. But if I had to alter it? That's not in my wheelhouse

Click image for larger version  Name:	DALL·E 2022-12-04 20.08.57 - Vintage horror children's book illustration.png Views:	0 Size:	18.5 KB ID:	5697330
 

Attachments

  • DALL·E 2022-12-04 20.08.57 - Vintage horror children's book illustration.png
    DALL·E 2022-12-04 20.08.57 - Vintage horror children's book illustration.png
    18.5 KB · Views: 0
Have you interacted with chatgpt? It's very informative about how AI works, where it excels and where its limitations are

It's too early...it doesn't know everything. Speaking about how AI works now, where it excels and where its limitations are is like speaking about the future of social media in 1995 by making educated guesses based on living life with AOL.

And that's only one part of it. What about the people/power part? All it takes is a Jeff Bezos or an Elon Musk to say or do something worthy with AI and there's an immediate butterfly effect (including completing destroying AI).

Someone does something successful with it and then suddenly every company wants in. Things could change quickly.
 
If you understand how something fundamentally works, well... same as with software generally. AI is software and shares the strengths/weaknesses of computers generally

Specifically, chatgpt works by computing statistical probabilities of which words are most likely to come next. That's why it can create naturalistic speech but it's also a recipe for unoriginality. Similarly, a lot of AI art feels like things I've seen a thousand times before but just lack the skills to create myself
 
ha...a lot of human work feels like I've seen it HUNDREDS of thousands of times before. Same gimbals with slow-motion, same intros with the flashy text describing what the channel does ("Reviews - Photography - Tutorials"), same dramatic establishing mood setter ("What's up YouTube!?!"...accompanied by a hand or head motion), same calling the audience a nickname and asking them to subscribe and "smash that notification bell", same talking head sets with big monitors in the background or off to the side with nice RGB lights all over the place...same, same, same.

Ask an AI robot to mimic human camera operators and be "creative" and it will put its camera in 60p or 120p and just walk around with its internal IBIS. lol
 
Yes, the run of the mill I'm talking about. But styles evolve and ppl at the leading edge of that evolution win. Even (or especially) in wedding photography. Most evolution is incremental, so it's happening constantly

Interesting thread discussing if chatgpt can replace junior software developers. Consensus seems to be that it's more like McDonald's making burgers faster: you end up increasing burger (software) production volume and needing more workers anyway to handle the parts (test and check) that the AI can't do

https://twitter.com/shashiwhocodes/s...IAtEkZU0iLrUSA
 
But is that a valid question? Are all junior software developers created equally? Does the industry they work in matter? The software they use? Anything else?

I don't know but just seems like a vague inquiry. (The one person in the comments asking if anyone still hires junior software developers is funny, ha.)

I don't know how those industries operate, but I would ask how many software developers do companies really need. Presumably larger companies need more than smaller companies, and presumably the day-to-day varies accordingly.

So I'm thinking not all software is working and powering something the same way...a retail website using AI to build might have more flexibility than a bank or a security system?

Not the best analogy but I'm thinking some software development will be able to benefit from automated pipelines for simpler coding which don't need 24/7 supervision and won't need more workers to check its work. So if AI messes something up, they can fix it in the morning - but in the meantime AI just coded for 15 hours after everyone went to happy hour.

I know it's just a generic question on Twitter mostly for the engagement but I would think the real answer would vary.

Many may surely get replaced but many may not because of what a company may particularly do, or its values, or the top brass (maybe all tech heads on a warpath against AI), etc.
 
Apparently there are already AI coding tools being used in the field (CoPilot and/or ghostwriter etc) which basically recall specific bits of code from user prompts, saving time. Basically the programmer decides on the broader structure of what they're trying to code, and the AI helps out with rote minutia once the coder reaches a section of commonly used code. I would think that over time, the AI might be able to handle larger and larger chunks of the minutia, but I don't know.

Coding is happening all the time. Adding new features, adapting to new protocols, fixing bugs, rethinking and redesigning
 
sounds like an ideal system to me; let the humans worry about more important matters like their lunch orders while the AI handles the small stuff, and occasionally Joe-Joe or Jane-Jane can step in and see what's up if AI does boo-boo
 
Hypothetically, if given proper templates, one should already be able to do an AI film by him/herself without really trying. It may look like a Saturday morning cartoon but it's better than nothing. A real pro ought to be make a really good looking cartoon though.
 
I think all of this recent fascination with AI in the past 24 months is interesting...

https://twitter.com/andrew_n_carr/status/1599979082350927872?s=20&t=RrqdkCdmURvDwXkqcifhQg

There have been a couple advances in usable AI that have caught the public's attention, mostly text-to-image stable diffusion, and also this chatgpt. The question is where it goes from here, but these 2 are sizable leaps forward already

Talking to chatgpt is so natural. I don't have to tailor my speech in any way. It can understand what I'm saying as if I were talking to a person. Colloquialisms, incomplete sentences, etc.
 
Back
Top