In a new Sony Corporate Report, Sony has revealed that PlayStation will use AI and machine learning to speed up its game development.
On page 16 of the report, Sony had that “bolstering technologies that can help creators engage in maximizing the value of their IP in efficient, high-quality ways, including sensing and capturing as well as real-time 3D processing, AI, and machine learning,” and that these technologies will help to deliver its IP “rapidly and at low cost to a broader range of fans.”
The report reveals that PlayStation used machine learning in the production of Marvel’s Spider-Man 2 by applying voice-recognition software in certain languages. This process allowed the company to automatically synchronize subtitles with each character’s lines to “significantly shortening the subtitling process.”
So back when we were waiting for Guild Wars 2 to release, a youtuber named WoodenPotatoes was invited to Arenanet to review the progress. He noticed in his videos that he saw several devs working on animations that were really not that important but would be immediately be noticed missing. His example was drawing and stowing weapons. With 5 races am a good dozen weapons this took an incredible amount of time. Now imagine that you can train an AI to do this and only have animators polish the result. A lot of time saved for more important stuff.
You think a game has not enough models or all the faces look alike? Not enough hair? Let an AI take care of that and have designers polish the result.
The forest doesn’t look organic and too constructed? Have an AI naturally grow the forest. Wait, there are plenty of games already doing that. When CEOs talk about speeding up game development they don’t mean to push out generic games fully developed by AI (well some might mean that) but to tackle the aspect of game development that slow the entire process down but not adding quality.
Given that Starfield took what? 8 years to development and resulted in a (according to the internet pretty bad) generic aged science fiction RPG. I’d prefer some AI supported development when the overall quality increases and AAA game development is not longer a decade long project.
There is plenty of bad things to say about AI but it does offer improvements.
AAA games don’t have a production quality or even a development time problem. They have a far more existential one. A gameplay focus problem. These are games made with profit as first priority, not fun. They have confused engagement and addiction with gameplay quality. Live services poisoned their design language. This is why they want more, faster, at higher budgets. The fallacy is that more, faster, more graphically demanding, will magically make them all the money.
I want less games, with lower budgets, that take longer to make, have less graphic and animation fidelity, that pay better to their devs to do their job well. And I mean it.
The video games market is already overflowed for its size, yet somehow these companies are inflating their budgets like balloons instead and charging ever more and more for shittier games that somehow cost more to make. This isn’t sustainable. AI won’t fix any of these issues.
I find myself buying almost exclusively from indie devs lately. They make games that still have a soul and aren’t driven by this stupid mentality that better looking magically equals more engaging gameplay.
There are probably a ton of devs in the video game world that were once passionate about making games, that have since been burned out by the industry’s grueling demands. AI is a bandage on a far bigger existential problem and that real problem is capitalism.
If I see a game that costs $70-100 now, I drive right past it. So many of those high dollar AAA turn out to be absolute duds that have live service and other BS jammed into them that some suits in a boardroom thought up.
If you truly break it down, you’ll notice that AAA only actually makes two or three games, open world third person action RPG with parkour, open world shooter with looting and crafting, and live-service coop/competitive shooter with loot boxes. Every iteration of these same ideas are just varnishing the same bored gameplay concepts over and over with different coats of theming and slightly different stories. I only ever find original and stimulating gameplay on indie projects and the occasional small studio. They’re the only ones actually experimenting with innovative game design and varied concepts.
Yeah, 100%. They don’t build games first; they build a profit framework and then build a game around that.
you can already do that with regular tools, you dont need ai bullshit for his
and ai writing would create even more generic bethesda rpg than any person could even dream of, are you huffing some techbro fumes?
Not huffing anything, strict non smoker here. I’m just working in a non games tech area and can see, how two ML OPs developers do the work of 10 engineers. Work that those engineers never wanted to do in the first place that took up the vast majority of their time and prevented them from doing their actual work, preventing catastrophic failures of large industrial factories.
We totally agree on AI writing but there is way more than this to AI.
And “AI bad” is en Par with “ThAt’s sOciaLiSm”. I’d rather reap the benefits and manage and regulate the risks of a new technology than to condemn it completely because I don’t like certain parts of it
For me personally, the solution I prefer to see for “Our idea for this game is shaping up to be packed full assets that will swamp development” is for them to find some excuse to cut the content. Genuinely. Artistry thrives in the presence of limitations.
What games out of curiosity? You don’t just mean normal procedural generation which has been around forever? It’s not the same as using AI to generate a million different haircuts.
The first game I saw with advanced logic was Far Cry 3. The devs would simulate how tree spread, die and wither and then let this run for hundreds of generations. Then they would alter the forest to fit in buildings und other stuff
The animation stuff you mentioned exists today:
https://www.youtube.com/watch?v=Gt1yNJ180Cs
https://www.youtube.com/watch?v=_AnovXOJOaw
https://www.youtube.com/watch?v=nAMSfmHuMOQ
this is the sane way to use AI, let it to mundane tasks so people can do more rewarding and fun stuff. unfortunately for most CEOs and shareholders, AI=how many employees can I fire and replace with AI.
Can you imagine them investing all that money in AI and the end result is their employees can work less but earn the same salary? yeesh they would rather die