AI making 'games'
(deepmind.google)
Comments (23)
sorted by:
I have to admit, it is VERY funny to me that for YEARS we've been hearing how 'all working class jobs will be made redundant with AI, all that will be left is the arts'
When what transpired is arts is the FIRST target of AI and working class jobs either are still untouched or improved productivity with AI. I'm starting to believe they were projecting their fears.
True enough, because logically speaking, if you can automate and use AI to do the job of 20 factory workers, you're still going to want those factory workers around for if the machine fucks up as that is always a danger and they have the experience to possibly improve the process with said machines.
The western artists have shown themselves how generic the slop they make that they can't even tell what a woman is so AI is already surpassing them.
That's a big part of it I think. And the Hollywood writer's strike tied into it. If someone is doing good work then even if the AI stuff is cheaper there's arguments to be made for keeping the person. But, a lot of the writers and artists these days just keep pitching misses (and wide misses at that) over and over. If you're already failing over and over then there's no excuse to not replace you with a machine that does the job faster, cheaper and better.
Pretty good idea for a comedy sketch. A writers room keeps pitching the same premise (black, lesbian, trans) and a head writer keeps shooting it down. The writers get increasingly agitated as pitch after pitch is shot down, even though they’re all the same.
The difference between AI and artists is control.
Doing it yourself gives you more control but AI is faster.
For now at least. AI looked very different last year and it will look different again this year. What you can find today is the worst it is ever gonna look from now on and AI art will probably be best used by artists who know their shit, yet it can also be used by amateurs to make something good looking. The people who cry about AI WILL be left behind sooner rather than later.
It's just useless at doing anything productive, so they're focusing on art which has no quantifiable metrics for success. It still sucks at that also. Just fewer suits which care that it sucks.
Not useless so much as less cost efficient in a lot of cases. Still very much useful in a lot of areas, but a lot of the potential use is still so experimental and new.
Not specifically because it's buggy/unoptimized so much as that the design and incorporation of it is so much newer and different than most technical innovations in the past decade or two.
Takes a bit more out of the box thinking than the usual "autism".
I see it as a tool to be used. In all honesty, most of the programs used today for design jobs have not had a big improvement in years. The designers are using them, but not really understanding them.
AI isn't a danger to their jobs, but the lack of improvements has been a danger this entire time. AI is just doing what needed to be done 15 years ago.
`Aye, that's been my observation as well. And I think the few designers who have some idea of how to really utilize it are not exactly eager to unveil the curtain too early, especially to potential competitors.
And I guess since a lot of it usually occurs behind the scenes of development, it's also not the sort of stuff that's usually worth showcasing to consumers.
There's a lot of smoke and mirrors with stuff like this. Bullshots are still a thing.
They absolutely are. As with all capital investment, the automation never eliminates the workforce, it just forces the workers into different positions and areas of responsibilities.
John Henry is an excellent example, because it's their claim that basically the industry would die without the miner. But that's not the case. A steam powered hammer drill killed a man because he could barely keep up with it. But what do we have now? This gigantic monster. People stopped hitting nails and instead took up over-seeing the machine, or doing other work in the industry until they retired. This includes training. If it was really so bad that a worker couldn't adapt, they could be fired, but would most likely just go and work at a smaller company that couldn't afford the new capital investment yet. Like a smaller rail road company that doesn't have the money for steam drills.
The working class who is good at their jobs are going to be just fine. They are still competitive and will get improved productivity.
But what of the bureaucrats? Well, if they can't manage any organizational structure, they are gone forever. They don't actually have a transferable skill if they are a bureaucrat that isn't actually good at bureaucracy.
That's the trick here. Even the bureaucrats, who are good at bureaucracy, will continue to exist and can transition out into other organizational logistical efforts. What you are seeing are the useless people who won't adapt, refuse to. They are the people who are losing their jobs and careers and that's a good thing.
And I don't just mean a subjective good from my political stance, but an objective good for everyone. They cost consumers money, they cost businesses money, and they are normally a problem in the work place. You should be firing these people. Twitter fired 85% of it's staff and suffered minimal disruption. Among my employers and my friends employers. Firing 10% of the staff could collapse the company. That was all bloat.
The bloaters are worried the AI will call them out.
Damn. The guys walking along that machine and especially the guy just sitting in it as concrete railroad ties are dropped have some balls.
But, yeah. Trim the fat. Tell them to learn to code or go mine. Or, if we are lucky enough, return Detroit to what it used to be and they can go stamp out auto parts. If they aren't willing to work such an "ignoble" or working class job, let them starve.
we are far from perfect ai.
AI makes an error in production / self driving car? Thousands of dollars down the drain, maybe human injuries/deaths.
AI makes an error when drawing a pic? everyone laughs at the wrong finger count and requeues.
that is btw also a reason why good artists don't have to fear AI especially when they integrate it into their work flow for stuff like faster prototyping.
I feel like this sentiment has come entirely from the management class because anyone who actually works in computers has had the "replace a person with a bash script" experience. I've been saying for over a decade now that the safest jobs will be anything that moves atoms in the real world because robotics is much harder than computer science. Bonus points for jobs that take place outside, on rough terrain, more than 50ft away from an electrical outlet, or outside of cell phone coverage.
AI for game logic is a fools errand. There's no way it could ever be reliable.
Graphics inferencing will totally be a thing though, possibly very soon. Render a basic scene, skip all the shader stuff and go straight to generating the graphics style you want. AI models have also proven to be very good at approximating light transport and particle simulations, so water and smoke and diffraction and transparency could all just happen for the the same linear time cost.
AI with interaction like this would be much more suited to VFX and creative applications where you don't want strict rules. Being able to adjust a scene with real time feedback would be very useful.
I’ve always thought generative AI could be best used in something like a sports game. Take college football the video game. I play NCAA 11, and each week of the season it would create headline blurbs out of the simmed results, using player stats, game stats and team/conference records. Using those stats, I think it would be cool for AI to generate a nice little article about each of the big games and the team you’re coaching in Dynasty mode. It would add a lot of immersion and also make it easier to keep up with the wider season as it is quite boring looking at stat spreadsheets after a while.
Scenes are good start. That's where I am starting at for my personal learning.
Using AI to procedurally generate environments is the wrong way to go, they should be focusing on NPC behaviour and dialogs. Though I suspect there would be a lot of resistance from voice actors, as they'd be the first ones to be made redundant.
I agree to a point. As long as the assets are build within the parameters of the art design established by a human, it should be just fine in simplifying the artists' efforts.
That's being worked on too.
Agreed. They have them now but it's really, really slow. In about four or five years it should be standard in a lot of games (hopefully).