EA CEO: "Real hunger" among developers to use AI to speed up development | Hacker News Reader

EA CEO: "Real hunger" among developers to use AI to speed up development

videogameschronicle.com

23 pointsjarsin32 comments

What EA is asking for is the ability to cut even more corners in their game development pipelines so that they can release a higher volume of unfinished titles, all of which require months of patches to be applied before any of them are reasonably playable.

Sounds great!

Don't forget about the game breaking DLCs that they'll gladly take money for but never fix.

If you think about what is fundamentally required for a game, in an ideal world, you don’t have any programmers at all—just artists and game designers.

> If you think about what is fundamentally required for a game, in an ideal world, you don’t have any programmers

In an ideal world, no. In the real world, your 3D surface must be displayed on a screen.

> just artists

Well, oops on that, I guess.

Ideal for whom, though?

I mean ideal as in the physics notion of an ideal spring, simplified from all of natures forces so it may be modeled in a useful way.

In the same way, an “ideal” game is one separated from its inner workings, and what you’re left with is the user experience, beautiful artwork, music, and engaging entertainment value.

I think this is missing the creative aspect of taking an idea and turning it into reality. Developers spend all day finding the middle ground between an idea that won’t work and a simple realistic solution that will

Sure and that’s a reality today. That process will still happen. We’re talking about AI’s doing that for us don’t forget. And those mechanics will still be driven by prompts and such, but the nitty gritty will be automated.

What a weird future.

AI reduces the need for humans, AI then makes products for people to buy. Those people now don't have jobs to consume the products made by AI.

It's almost like we need a new economic system which guarantees a nice lifestyle even if you just don't want to work. And then if you want nicer things, like a bigger house you can be one of the lucky ones who can still find a job.

It's not to far off to say within 20 years driving/transportation will be completely automated. We'll have mail robots drop off packages. Robots building houses, etc.

I don't see a utopian no work future though. More like a dystopian high income disparity future

I feel like I'm noticing this more and more often in media in general and definitely hacker-news-circulated media.

This weird future is straight out of manna - https://marshallbrain.com/manna1

I've read that short story before.

It's wildly optimistic to think the government would opt to give everyone housing over just letting them struggle.

The future involves UBI - there is no other way.

But, before that,

- A lot of people will be shed from jobs because companies seem to prefer this over loss of shareholder value

- A lot of products will become shitty as they try to extract money.

- A lot of companies will collapse as people will not have the money until UBI is implemented.

Thanks for seeing my bigger point, I don't know the specifics of course but less people will have jobs in the future.

As is UBI would probably be more efficient then the patchwork of programs we have now. But I don't really see that happening anytime soon.

I'm not sure what region you're talking about but the world is really huge and varied, in 20 years no way every country will be at that stage of automation.

Seems like the difference between cell phones and landlines implies things like self driving cars will be spread without relying on most people in a region affording them before construction starts.

The fully automated vehicles may be a lot more dangerous in some places than others, but you'll probably be able to get run down in the middle of nowhere.

>We'll have mail robots drop off packages. Robots building houses, etc.

Robots building houses? I don't believe it

Robots/Machines need not small maintenance squad in great environment like semiconductor fab, but somehow they'll manage to be productive in mess like construction site?

the precision required in home building is incredibly low, a drunk human can frame without even thinking.

It will probably just be a small group of people maintaining a few machines doing the repetitive work while humans put in the final touches and fix the broken shit.

definitely 20-30 years out though, its so much easier said than done.

precision maybe is not at the semiconductor level

but there's shitton of tasks of varying kind to do where each of them can/is usually customized (unlike manufacturing line)

They don't have to be at a construction site. It can be in a factory on a production line. Prefabricated homes have been around for a while. Mobile homes are probably the most common, but there are also modular and panel homes that can be made in a factory, then put together on-site.

This site has good info about factory homes. There's a lot of cool things going on in that area.

https://modernprefabs.com/

AI reduces the need for low skill humans, the skill ceiling is just being raised. Even AGI won't replace most jobs because literally nobody trusts it, and the companies who do will eventually suffer fallout when it fails. It's just like self driving cars, nobody cares when humans crash, but if a self driving car slows down traffic people are burning them down in the streets.

The only people who are fucked are people in third world countries because AI is cheaper than human labor. Inequality is gonna go through the roof and if I lived in a crappy place I would be doing anything I can to get to a first world country.

>AI reduces the need for low skill humans

Except the focus on AI has been replacing _high_ skill jobs like artists, writers, and developers. The end-goal of AI devotees is that humans will dig the ditches, and AI will do all the 'fun stuff'.

It's not a great plan, but they seem really gung-ho about it.

Close - "AI" is replace high paying moderate skill jobs. Those artists, writers and developers that are being replaced are not in the innovative and creative roles, they are just the assembly workers putting together or "fluffing up" work of other people. Not great for those people and increasing the inequality overall.

Real “hunger” among the MBAs to replace all the actually skilled employees with low paid code monkeys so that mbas can make even more money.

That’s the MBA model of business: don’t create anything of value just try to extract as much of the existing value as possible while short changing anyone who actually does real work.

More like corporate has been screaming about ai And wannabe leaders keep saying yes

> “As a company, we’ve been deeply tied to AI since our inception,”

This kind of statement is just so weird. Really? They were deeply tied to AI in the 80s? When they were launching their first games and calling devs “software artists”?

Unless they’re talking about computer players and NPCs, but… what a stretch that is. It’s like greenwashing, it’s… AI-washing. Very silly.

As a game AI developer since the 90s, I can say that lost and confused recruiters have only recently started tying me to machine learning AI, lol.

Game AI is closer to robotics (“perceive-think-actuate”) than machine learning. Always has been and continues to be. The think part is achieved by anything from a script to a state machine like behavior trees, to utility or planning networks. But it would be woefully inefficient to think with neural nets.

So the two AI are very unrelated. Neural nets are used in games only a little bit. Ubisoft had a research division where they made them play animations and drive a car in game. Epic Games built a ML deformer for character model meshes, which is faster than kinematically resolving realistic deformation (skin stretching). In mobile text based games, there was a time of experimentation a bit earlier. But game AI is much more complex in the breadth of types of calculations that it does to benefit from slow and imprecise generative AI.

Offline, yeah, game developers want to use Copilot and Dall-E to create filler content quicker, whether boilerplate code or textures for environment art like paintings, lore inspectables and decals. Also placeholder TTS has been done with generative AI for a while (like Google/Azure/Amazon TTS services). People are doing these things, minus Copilot probably, in the AAA space. Then the assets get replaced by final ones before shipping the game. This is great, it allows prototypes to be closer to final quality, which makes the game more cohesive.

But EA’s CEO… just wants to pump the stock to sell like most tech CEOs, in my humble personal opinion.

To recap: game runtime AI is AI but materially different and the overlap is probably through robotics more than directly.

Generative content in games has been around at least since Nethack? "AI" in games has been around since Pong.

In pong, the bar literally just moved towards the ball at a set speed depending on the difficulty. That is not AI

"AI" has lost all meaning years ago, it's just marketing fluff now and a great way to pump stocks.

On the one hand, gaming is one of the few industries where generative AI can have a big impact on producitivity today. It's easy to be cynical, but used well it could reverse the trend of excessively expensive AAA game development.

On the other hand, who would trust the CEO of EA / EA in general not to do all the wrong things with AI to save a buck.

TL;DR: Generative could help a lot, but big publishers will use it the worse ways possible.

I don't trust EA, no, or many of the large publishers. I do, however, trust many small- and medium-sized publishers, and particularly indie devs/publishers.

In short: I trust the industry at large to adapt properly.

(As others have pointed out, there will be growing pains)

Reply on news.ycombinator.com