2024.07.05 While the gaming industry explores the potential of generative AI, Nintendo remains cautious due to concerns over IP rights and the company's predilection for its unique approach to game development.
Anyone know offhand their stance on jackbooted thugs kicking in the doors of people who write emulation software and sending them to prison? Just trying to get a pulse on that
Nintendo is probably the most anti emulation company. They tried to make emulation completely illegal, they sued yuzu (the devs did some things that encouraged that) and made the devs pay a couple million and never develop emulators again. They are also the reason why dolphin didn’t come to steam as they wanted to sue valve if they allowed it on their platform. They also took down various tools that allowed one to rip your own switch games. I think they are for the second point
In this interview he claims he was simply paid to develop like a contractor and the people running the business still haven't faced consequences: https://darknetdiaries.com/episode/136/
I mean, it feels kinda obvious. If there's any company in this space that would never need to use generative AI, it's Nintendo. They already employ some of the most talented art teams in the industry, they're not exactly struggling to produce art or assets.
If this was from a developer who maybe hasn't been a gaming monolith for the last 30+ years, that'd be different. This is like if Bill Gates says he promises not to open his 401k early; like, okay cool, I don't think that was ever in doubt in the last 30 years, Bill, but thanks for letting us know.
Activision, Ubisoft, and EA, all multibillion game dev company, said they'll be using generative AI to make their game, so no, it's not really obvious. It's also mentioned in the article
Nintendo's stance differs from that of other gaming giants. Earlier this year, Ubisoft introduced Project Neural Nexus NEO NPCs, which uses generative AI to simulate in-game conversations and interactions with NPCs.
Similarly, Square Enix President Takashi Kiryu views generative AI as a business opportunity to create new content using cutting-edge technologies. Electronic Arts (EA) has also embraced generative AI, with CEO Andrew Wilson predicting that over half of EA's development processes will benefit from advancements in generative AI.
The investor probably heard it's a trendy thing that can make better profit, so they asked nintendo about it in the briefing.
Those are the companies that have contributed to the fast churn of creatives getting overworked and leaving the industry, leaving their projects to be driven entirely by excess man-hours and lack of innovation.
We haven't really seen high quality art that uses AI as part of the creative process yet, but this could be similar to the animation studios of the 90s who refused to use computers. They're all out of business now.
The reality is, generative AI is a really powerful tool, so they will be at a disadvantage going forward if they don't use it.
Instead of generative AI for game assets, id much rather see something like a LLM in game that dynamically controls NPC behavior. That would be cool as hell.
Like an RPG where you can type what you want to say to an NPC instead of choosing a fixed dialogue tree.
Only if they use it the same way and within the same context.
But isn't that what always happens when a new gaming system/idea explodes and clones start poping up?
I don't think that matters much, in fact competition might actually be a good thing.
So far research suggests the guardrail and hallucination problems are unsolvable, and we are seeing diminishing returns from increasing the complexity of these systems.
Hence devs will never have the necessary control required to author an actual narrative. NPCs will end up talking about mechanics that don't exist, or saying things that contradict an overrall narrative.
Even with actual people, if you just throw them in a room and have the improv a world into existence, it never ends up quite as good as a properly authored narrative.
And LLMs are nowhere near achieving the level of internal consistency required for something like the worlds of Elden Ring or Mass Effect.
Baldur's Gate 3 contains truly staggering amounts of writing, multiple times that of classical literary works. The hallucination problem means that if all that were AI generated, small parts of it might pass inspection, but trying to immerse yourself in it as a fictional world would have you noticing immersion breaking continuity errors left and right.
They mentioned IP right, and since Japanese police do arrest and charge people for infringing copyright, and since some artist do recognize their work being smashed together to create something else, they might skip this thing for the foreseeable future just to be on the safe side. They know because they're the one that send police to arrest people.
The article talks about how they are ok with using AI for things outside generating images, texts and so. For example, they are fine using the rudimentary AI of any typical enemy in one of their games. So I expect procedural generation that does not rely on trained bayesian network models is ok for them.
It looks like they just seem to be concerned about the legality of it... so they might just start using it as soon as the legal situation for AI models is made safe.