Sakurai: modern game development is too time consuming and unsustainable. AI is the solution.

This is exactly the direction of travel the last few years. Shit is not sustainable and they see AI as the answer. When they should be looking in the mirror and asking themselves if budgets and scope etc need to be as big as they are? Or even easier ask themselves two questions:
1. Does this really need to be open world?
2. Does this really need to be a live service game?
 
If you want to blame it on cultural issues, okay, but I have a follow up question: Why are asian-region devs suffering from the same issues?
Huh. Thats funny. I would've thought you'd imply that they aren't suffering the same problems.

But you're right, there are to a large degree. It isn't because they're chasing cutting edge graphics, as you know. Metaphor took 8 years and it's barely distinguishable from last gen Persona 5 in terms of asset quality or design tenets, for example.

I think they're mostly having the same problems as the west, and developed having those issues in the 7th gen of consoles. It's easier to find the quotes, articles, tweets, and videos from western game devs to confirm their issues, but I do think it's easy enough to tell that the Japanese sphere - which maybe has an established senior layer that is even more entrenched and worshipped than the western side - is dealing with the same stuff.

If you want a case study, it would probably be best to look at Square Enix. Their production processes for AAAs has been inefficient for the last 20 years.
 
This is exactly the direction of travel the last few years. Shit is not sustainable and they see AI as the answer. When they should be looking in the mirror and asking themselves if budgets and scope etc need to be as big as they are? Or even easier ask themselves two questions:
1. Does this really need to be open world?
2. Does this really need to be a live service game?

yep. sometimes, the cheaper solution & the better solution aren't always the same solution...
 
Who said games must always be bigger? It's either Todd Howard or Take Two's biggest head, I can't think of another one.
Funnily enough I think it was Oblivion that started the trend back in the day and then it just spread around.

But the problem is really that not enough people are willing to pay full price for a short game. Simple as that. Every now and then some game slip through the cracks like Space Marine 2 last year, but overall the risk of failure is probably too high.
 
How was he saying it?

Was it ~"I'm using AI, get fucked, Kirby with 6 fingers!!"? Or "If people want to keep improving the realism they'd need to find ways to cut down the cost and time using some AI on the simpler tasks?
 
Sakurai's approach to modern game development, which involves such practices as giving himself IV fluids so he can keep working while terribly ill, is absolutely unsustainable.

More reasonable and self-preserving approaches already exist without the need for AI.
 
It's kind of scary hearing it from him specifically tbh, knowing the workaholic he is.
Donald Trump Politics GIF by BuzzFeed

Exactly, also makes many of the replies here seem idiotic. If Sakurai says this, everyone should be listening.
 
I've been saying this here for a long time now, the ideal future would be AAA devs using the help of AI to get shorter dev times and AA and solo devs could use AI to reach quality levels that would be pretty much impossible for them in the past years. That is my hope, of course there will be tons of slop but creative people will rise i think.
Also there is an even bigger "AI future" that could render all this discussion pointless if generative AI video gets good enough and fast enough that you can infinitely explore it in real time, instant infinite explorable worlds at the click of a button. That would change thinks quite a lot but i still predict that traditional gaming with human crafted games (using AI as a tool or not) will keep coming out.
 
It's pretty interesting to hear that from someone whose career is tied to Nintendo. That's one of few companies right now that doesn't have to worry about the development process of their games. Nintendo's top games don't have 200 mil+ budgets and they don't have to sell millions of copies to break even... and they still do that, because they have a very dedicated fanbase.

AI isn't a magical solution to the industry's problems, just like upscaling isn't a solution to poor optimization. It's a combination of different factors - there are too many games coming out, budgets and projects are often poorly managed, some games are too big and they take too long to develop (risking the fact that current trends may change) and there's an increasingly aggressive chase after monetization.
 
5+ year dev cycles are insane (with very few guaranteed-to-succeed exceptions), and not just because of the expense of developing for so long but because of how much the market can shift in the interim. Most studios need to stay nimble and be able to respond to changing trends; being locked into a project for so long makes that impossible. This is how studios end up doing stupid shit like developing a Wonder Woman game for multiple years beyond the point at which it was guaranteed to flop, or having to delay a game for months to minimise the woke content they already spent months putting into the game.

If AI makes it possible to get back to a far more reasonable ~2 year cycle, without sacrificing quality, it will be hugely beneficial to studios. Beneficial in terms of it being more likely that the target will still be in the same place by the time they get to release, and in terms of making it more viable to take risks to begin with if it's taking much less time to reach a point where they can know if that 'riskier idea' is working out or not.

There are also some things which AI can potentially make possible in gaming which we currently just don't get because they would be prohibitively expensive with a human workforce. These possibilities don't really get enough consideration, but that is a slightly different topic.
 
just make shorter games like on xbox 360/ps3.
I would prefer 8-20h games over 50-100
Do you remember the obsession people had with 'hours of gameplay' during PS3 gen? EVery single linear game had some variation of 'short hours of gameplay' by professional reviews which would bring down the score.

Consumers were obsessed as well (to an extent still are) whether it provided 'bang for your buck'
 
Do you remember the obsession people had with 'hours of gameplay' during PS3 gen? EVery single linear game had some variation of 'short hours of gameplay' by professional reviews which would bring down the score.

Consumers were obsessed as well (to an extent still are) whether it provided 'bang for your buck'
He is the problem. Same with order 1886. Is the poors justifying their money
 
Why did it become time-consuming when engines have evolved considerably to develop games easily like UE5(even tho it's not a great performative engine)?
It still needs a tone of work and optimisations. UE gives people a lot of functionality and makes ideation process easier and faster. In the end thought programmers and technical staff need to write a lot of code and make proper optimisations. That's why many UE games made by smaller studios aren't on the same level (visually or performance) as those from bigger companies (well, it's one of the reasons why).

Corporations used to make a big game every 2 year or so and those game were harder to make because artists had to fake realistic lighting and whatnot. Now with raytracing it's much easier. Sounds like bs.
Well, lighting is only one of the many parts that makes a game. But even if we'll stick to that - do we have RT in every game? In games where there is RT, do we have a full package, or only parts of it? Are those games are using RT only or have a older solutions as fall back for GPU's that don't have RT acceleration (or are just weaker). You can say that dev's have now more work to be done - you need to make lighting the old way and the new one in one game. Usually it means that RT is made with lesser effort.

I think it was still too early for RT implementation. Nvidia forced it and it became more popular and some gamers can enjoy it, sure. But I think that until you can't have it in the cheapest GPU's with good results (so like 'for everyone'), then it's not worth it. Consoles can push that better, but the technology is too expensive and frankly still not good enough for dev's to use it freely like, let's say normal maps.

Some users here are writing that dev's should go back to PS4 era visuals. Let's take characters for example. The way we are making models for PS4 and PS5 is the same. We are sometimes pushing the high poly meshes more, but usually this transition means that we can bake higher resolution textures. We can be less picky about polygon budgets. But the time that we spend on making those character models didn't changed much during those 12 years. We still have more or less the same time budgets. The process is a bit more complicated (like cloth simulation, hairs from splines, etc.) but the same time it can be a bit less in other departments.

I would argue that we would need to get back to PS3/X360 generation for better time savings. Or better, to PS2 - this is where we could make models faster. On the other hand less complicated work would mean that we would need to pay attention of other things (but still it sould be faster).

Anyway the main issue is time - it cost's money. Through years our salaries went up, because the cost of living went up too. From my perspective - I'm earning money that I wasn't imaganing when I started my carear. Do I feel it in my life, especially through last eight yeras? No, I feel that I'm still on the same level but I earn more then I was. You can't do anything about it, not in game dev at least. Well, you can fire most of employees and replace them with an AI. Some day at least ;).
 
lol, lmao even.

Isnt this the guy that is constantly crying about how stressed he is and how terrible making video games is?

Yeah, fuck off. AI is there to speed some process, and I can understand and support it: like just giving all the documentation of a software to it so your devs can solve problems and questions quicker, or have some sugestions (that have like 50/50 percent of being wrong btw. Thats when the dev experience is important). But Assets, textures, music and others? I dont think so.

Games DO need to scale down. Idk where they are putting the money, but all these corporations are too bloated and games not necessarily need many thing they put now.
 
If you want a case study, it would probably be best to look at Square Enix. Their production processes for AAAs has been inefficient for the last 20 years.
The problem of SE is that their core games are JRPG meant for local market. That essentially dead making it very difficult for SE to adapt.

Games DO need to scale down. Idk where they are putting the money, but all these corporations are too bloated and games not necessarily need many thing they put now.
Market doesn't really want it
And it will be even harder to justify 80$ price point, especially compared to F2P games, those continue upping their value proposition being free

It's not like there are no smaller games with simpler graphics. It's just 99% are unknown and left to rot somewhere deep in Steam library
 
I hope he's just dangling that while preparing a full-blown Kid Icarus: Uprising remake for the Switch 2.
He is not doing that, if Nintendo wants to remake that game they don't even need him he already made the game, that game sold 1 million copies no one is remaking it.
 
Honestly it makes the most sense that AI could help cut corners and shorten the time spans of development. I feel like implementing AI to do such things would make the most sense. Much like people using it now to help with mundane tasks. Hell, much better than anything art related, lol.
 
Top Bottom