Smash Bros. Director Sakurai Laments 'Unsustainable' AAA Game Dev, Discusses AI And Uncertainty

My body is ready for Kirby Air Riders!

I'm certainly not looking forward to an AI filled future.

These days, most AAA games — like The Last of Us, Until Dawn, Uncharted, and even Horizon to some extent — feel more like blueprints for future film franchises than standalone games. Personally, I find more enjoyment in smaller titles like Spyro or Ratchet & Clank, which are, at best, considered AA games.

Thanks for your professionalism Gavin Lane. True journalism right here!

Thank you for actually trying to convey that his view isn't as clear cut as people may make it to be.
It's a lot more neutral than it sounds.

Then make them with smaller teams and less copypaste content. The best games in history were made with much smaller teams. Not every game needs to have 100000000000000 polygons on a trash can.

Also, how is it complex if there is currently no budget for story and writing as it's all written in a tram on the last day?

Same with music - western games dont even have it anymore.

That's the reason western games are so far behind Japan, with no fighting chance.

As an AI sceptic I do believe a lot of people get alarmed for potentially no reason. There are applications in game development that can benefit from using AI (for example, generating asset variants) and have done so for a long time. On the other hand, using AI to generate valid code or even whole games.. good luck with that vibe coder. May the odds forever be in your favor, although I doubt it.

@Olliemar28
Oh, don't worry.
My 3rd party kids games choice don't even have AI.
I already know the eshop games with AI cartoon image
That was very obvious of using AI and I will not touching games like that.
Looked scummy.

@Olliemar28 Sadly I think kids game would be more susceptible to that, since kids probably won’t notice/care and same for the parents.

Do companies nowadays have no managers? It seems like its either "we make extremely expensive game" or "we use AI".

In the 90s and 2000s teams were tiny yet they were able to create amazing experiences without AI. Tomb Raider was made by couple people and it changed industry, it was that good.

Stop making excuses and hire managers that can actually manage.

@Olliemar28 @Anti-Matter I wouldn't be surprised if those games will actually use AI more then AAA. Those games are made to make fast cash, and the less they cost the better it is.

Nintendo is slowly trying to cover all bases. Make games $80, without dubbing because voice actors are expensive, now AI.

Fun fact, Japan devs are probably least paid devs in the world currently so where is the money going?

I'm less concerned with AAA projects turning to AI tools for help than I am with their further acceleration of the shovelware landslide that's already overtaken pretty much every digital store. The big companies that abuse the technology will end up digging their own graves, but the small teams that try to do things the right way are in ever-greater danger of being buried under the swarms of opportunistic parasites pumping out soulless garbage. The Switch eshop is practically a lost cause already, so I genuinely hope Nintendo at least keeps up their reported strategy of firmly restricting what's allowed into the Switch 2 library.

Never had a problem with reusing assets. It was that which gave us an additional three standalone Sims games on PC via the "Stories" series in the mid 2000s, and no less than four additional GTA games on the PS2.

I don't get why we're seeing fewer such derivative games as it was a way to produce more content while developers work on the A project in the background. Surely that would have been preferable to waiting 13 years for a single new instalment in the GTA series.

But to hell with AI. Just last night I caught a new music video from a veteran singer who released an all-AI video with absolutely no artistic justification for it. It's sick and insulting.

Taking the easy way out will never compare to that which has had care put into it.

Plus, without knowing precisely where AI algorithms are sourcing their assets, major companies can quickly find themselves inundated with legal issues if others can make copyright claims over AI-generated material.

@JohnnyMind
Its just a sign of really bad managers, i worked in couple huge corporations so i can tell there is massive difference between good and bad manager.

Good manager -> takes budget, takes number of people and creates space for projects that would fit into set budget while calculating all kinds of risks so they can be included in budget too.
There are always so many options in software development, in gaming you can just create smaller but more games.

Bad manager -> just "go with the flow" in other words "do nothing" and always shifts blame on external factors.

AI is just another tool that bad managers can use to hide their incompetence behind of.

I never liked the “bigger better” mentality in the first place.

Scaling back and focusing on quality over quantity would be welcomed by me.

Oh yeah? Then how come his name's Masahiro SakurAI? He's working on Kirby AIr Riders too.

Makes you think...

@RupeeClock It's kind of like how I view the whole debate surrounding AI and art. The way I see it, AI can be used to help conceptualize something, such an environment or a character, but developing that concept further and bringing it to life can and should be left up to humans.

For example, I created a couple of characters using Apple's new Playground app, which uses generative AI to produce images. One of them was an attempt to recreate Krystal from Star Fox. Mind you, this was all just for fun; I have no intention of commercializing the characters I created. But if I wanted to flesh these characters out further, I'd have to rely on a human to do so.

In fact, when I spoke about this very issue to my cousin, who is a professional artist, a week ago at a family reunion, she surprisingly agreed with me about how AI can be used as a tool to conceptualize things in artwork, just like how Adobe programs like Photoshop and Illustrator became a vital tool for artists (despite being a computer program instead of relying on traditional tools and methods; in fact, there was and still is a debate about whether art created in programs like those are considered "real art," given its digital nature).

Basically, AI should not entirely replace humans in the creative process, but instead supplement the tools that humans have at their disposal to be creative, just like Photoshop has been for the last several decades. People are, of course, free to disagree with this opinion. I always welcome debate on the subject.

There’s definitely more nuance going on here than “just throw it and AI and job’s a good ‘un”.

What’s certain is that the current state of large-scale game development is unsustainable.

“The only thing I can think of that could possibly break the deadlock is improved efficiency brought about by generative AI” …seems to be Sakurai’s general sentiment.

What’s not stated is exactly how AI is to be used. I don’t think Sakurai is advocating for the wholesale axing of game developers in favour of serving up fully AI-generated slop to unwitting consumers. But it’s undeniably powerful tech with the potential to both speed up busywork — resulting in a smoother production pipeline — or to create a passable but soulless pastiche of the product it’s purporting to be.

We also don’t know exactly what AI will look like in the coming years. Perhaps it will choke on its own output, growing dumber and dumber as more and more of its training data comes from AI-generated sources — like a form of technological inbreeding. Perhaps we’ll see the opposite feedback loop, where the technology can autonomously self-improve without external input, leading to exponential increases in power and unimaginable consequences.

The only thing certain about the future is that it’s anything but.

I find it funny a man who's friends with Kojima would be concerned about sustainablibilty of the industry, when Kojima spends most of his days finding NEW WAYS to waste Sony's money just so he can hangout with celebrities.

Which just makes Revengance, as the Metal Gear game with the least involvement form Kojima since 2 NES, all the more appropriate as time goes on.

AI really doesn't have to be a bad thing in some cases. Especially when it can help to make work more efficient. But what I'm opposed to is when it's used to replace work people should be doing, like making artwork, or writing good (manu)scripts.

I agree that the sustainability of game development is indeed under a lot of pressure. But let's not forget that one of the main reasons this is the case is shareholders and their bank accounts. Money and greed are what are truly ruining game development, especially in the case of these so called triple A games.

@AstroTheGamosian Yeah I'm fairly dubious of AI. Like if you use AI to generate dialogue for an NPC in my eyes there's zero point in speaking with that NPC.

Like why would I bother reading something that nobody bothered writing?

Here’s my idea:
1. No game should take more than 24 months to produce, unless there’s a really really good reason (like the game is truly groundbreaking) in which case you may extend the production to 36 months.
2. If at any point in production developers/publishers realize that their game will not break even if it sells “only” a million units, then rescope because this game is too big and expensive.
3. When a game is finished and released they should move on to the next game or sequel, not spend months and months fixing - sorry “supporting” - it. If a game turns out to be a massive hit they may invest some in dlc but this should be decided on after the game is already a proven success, not committed to in advance.

All studios of any size should follow these rules and they will be much less likely to fail after one flop, etc.

The fact that this sounds crazy and impossible is the problem in a nutshell.

@MSaturn
I remember when I was reading interview with Icewind Dale 2 creator. Managers gave him 9 months to make this game. You add maybe 3 months for "things we couldnt anticipate" and you get 12 months.

We went from 1 year of development cycles to 7+ years now. To anyone who has a brain this is completely crazy, It just shouldnt be allowed because once you cross 2 years mark you simply loose control of both finances and project scope.

@Dr_Lugae
I actually think AI enhancements to NPC dialogue is one of the few great uses of AI in video games.

Imagine you’re playing elder scrolls VII, and for the sake of argument let’s pretend the world hasn’t ended by then, and you are on a quest but you’re stuck so you go to the nearest village, go to the tavern and start asking around. But instead of choosing from a predetermined list of dialogue choices you actually type your real question into a prompt and then the npcs give intelligent answers. So instead of “mysterious tomb” from a dialogue tree and hearing the sane basic information word for word that the last npc told you you’re now able to say something “I know there’s a mysterious tomb near here but when I went to where it was supposed to be I couldn’t find it.” And the NPC could respond like “if you approached from the east the entrance may have been hidden behind the rocks. Try circling around from the west”.

I can’t wait for stuff like this. I hope the developers are smart enough to do it right.

@Dr_Lugae
Well some people use AI to simulate girlfriend now and they are happy, so yeah, idiocracy is real.

AI can be perfectly good for things like generating some NPC dialogs or whatever but the problem is that nobody trust these companies enough that they will not go "all in" with AI and start generating literally everything.

These companies are lika aligators, you give them your finger and they will eat you alive.

@slytherin_real Exactly. And there’s turnover, and target specs start to age and become outdated, and character designs go out of fashion and you wind up redoing work and everything spirals absurdly.

Common sense says if you’re going to spend 7x the time you need to charge 7x the price or sell 7x the units and since neither of those are realistic you get live service micro transactions as the “best” plan to try and extract more money out of users. It makes a certain kind of sense but it turns every game into a huge gamble.

@MSaturn
Not only a huge gamble, its just financially irresposible. Its like borrowing money, you will pay less borrowing 5$ for a year every year than borrowing 25$ for 5 years.

If IBM is successful with their SuperComputer, it will obliterate AI over night. If IBM fails someone else will succeed. Unfortunately it is the future.

I am not endorsing AI in this post.

I do feel like people see generative AI as mostly “slop”. It’s usually pretty easy to pick what AI did, and it really brings the product down when that’s what you notice. What people forget though, is that the generative AI stuff, for a big part, is still in infancy. If you look at let’s say pictures produced by AI, the progress towards realism in the last few let’s say 5 years is significant. In another 5, or 10, 15, etc. it’s only going to get more and more difficult to distinguish.

Ask yourself: is it the consumer demanding 4K, 100 hour games? I doubt it’s the publishers. In my opinion it’s the developers that are making things hard for themselves by setting out too lofty goals.

Something has to give. AAA games are massive and expensive and yet the market still demands more and more from them. Plus a game failure can destroy a studio. That is a tightrope that devs dedicate years of their lives to.

Personally I would prefer to see games reduce in scale (but I play a lot of games in different genres and budgets so I have tons of smaller games that I prefer to AAA games.) and. I feel that they have tools to make the job easier/faster. AI is one of those tools.

I will die on this hill… STOP… TRANSFORMING… EVERYTHING… INTO … OPEN… WORLDS!!
It’s just a damn trend that raises the development time and costs so damn much!

I’d rather have variety and genres.
I agree that making bigger games out of what once was a small/mid game is killing the industry.

Focus on quality… not quantity. Or at least dial it a bit down… there’s really no point in making games so overly massive that they become repetitive.

And just using AI will make everything so damn dull

@Summer235 Yeah, I mean kids like Roblox. One of the scummiest pay to play systems

@prw_88 Is it a quantum computer or something?

@Yoshi3 Couldn't agree more. Heh, maybe I agree too much, with my fondness for arcade games dictating much of what's in my Switch library.

But in the end, whether or not a game can sustain an audience beyond the initial purchase comes down to game feel and fun mechanics, which Nintendo swears by. The Switch sold like hotcakes, and the Switch 2, despite my grave reservations about pricing, is off to a great start-- and neither of those are dependent purely on open-world Zelda sales. (Or open-world Mario Kart sales, but come on, it's a new Mario Kart either way!)

And Zelda already had a long history of strong mechanics to put into an open world. When will AAA publishers and developers take a freaking hint?

If everything is open-world just for the sake of being "SO F**KIN' BIG," that's a good way to collapse the industry. Not even AI will save you, because then the customers will notice the game sucks. 🙄

@Antraxx777 yes. At least this is what my wife's uncle tells me. He was an engineer at IBM. He just recently retired.

quality > quantity.
human creativity > generative AI.

in the battle for the future of gaming, the people that know how to make games will win.

@Yoshi3
I personally like the games with huge world size to explore as long they are still for kids and have some activities to do when I play the game just for enjoy the random activities from wandering around, doing the available small activities like fishing or collecting fruits or diving under the sea, enjoy the scenery.
There are two different category of big world exploration games.
1. Free Roaming = Smaller size of Open world style games with limited area size to explore such as Super Mario Odyssey, Sly Cooper 2 / 3, Go Vacation, Paw Patrol World, etc.
2. Open World = Bigger and seamless size area to explore like LEGO CITY UNDERCOVER, Yonder The Cloud Catcher Chronicles, etc.
I will die in happiness to play new Animal Crossing with Free Roaming or Open World style for the world exploration + The SIMS style for designing the house and landscape.

@Anti-Matter You make a good point. Open-world, nonlinear games themselves are not the problem-- it's the AAA mentality behind their production in the industry at large. Not everything can be Skyrim. Not everything should be Skyrim.

If the aspirations of these games were realistic or intelligently contained and the visuals were more stylized, maybe these publishers wouldn't be laying off so many people a year.

@Anti-Matter I don’t think I agree. By transforming Animal Crossing into an open world you’re destroying part of its essence…. Creating your own stylized island landscape.

And that’s exactly my problem with trying to make everything open world, in the process you destroy the other elements that made the games before great.

It happened with Zelda, it happened with Mario Kart, it would happen with Animal Crossing.

Bigger isn’t always better. By focusing on the macro, you lose sight of the micro.

@Yoshi3
My personal opinion because I didn't grow by GameCube games so for Animal Crossing case, I don't even like the gameplay style of GameCube version by forcing you to play everyday or I will lose my villagers and the game was lack of customizing.
So when ACNH came with Terraforming and able to keep my villagers forever whenever I getting hiatus, that made me suddenly love ACNH more than ACNL on 3DS.
And after playing Dragon Quest Builders 2 with Minecraft style and Free Roaming in huge 3D worlds, I felt next Animal Crossing should have feature like Dragon Quest Builders.
Make it focus on designing and customization of characters, island, furnitures, etc + Free Roaming in big 3D Worlds with different environments + more side activities to do and I will love this new Animal Crossing concept even more than ACNH.
I'm totally happy to NOT experiencing with archaic Animal Crossing from GameCube style.
Just remove that lame idea of playing the game, replace it with something better.

Btw, Super Mario Odyssey was already the example of Free Roaming 3D World platformer and I like it better than Mario Galaxy or 2D Mario.
I have played another Free Roaming games like Sly Cooper 2/3 , Go Vacation with limited size of big area environments and I didn't feel lack of the gameplay essence.
I like to revisit the vacation resort in Go Vacation especially the Beach just for wandering around in a big world, enjoy the scenery or participate in sport activities if I want to.
And let's not forget I also have Zelda clone game Oceanhorn 2 on PS5 with Free Roaming / Open World style and the gameplay still felt similar as Zelda games even I have less interest with Zelda games.

I also played LEGO CITY UNDERCOVER for 3 times (Wii U, Switch, PS4) with seamless Open World style on New York & California style LEGO World and I enjoy to explore those whole worlds for free Roaming, having fun destroying the objects to collect the LEGO bricks, finishing the side quest when I'm in the mood.
The gameplay didn't ruined just because it's Open World.
Open World games can be great as long the developers know how to make the free Roaming activities not boring.
I'm definitely will wait for another Free Roaming / Open World 3D Mario games with design level like Mario movie 2023 style + voiceover + ability to play as other character.

I have to agree with Sakurai-san.
I've seen games go from enjoyable and affordable affairs with dev cycles of just a few years, to massive games with loooong dev cycles, too often in development hell, that're more bulky than is particularly good, and cost an arm and a leg!