While I think the current technology couldn’t do a big things like character/weapon/zone design anywhere close to the level of a flesh and blood person can…
I do think what AI would absolutely excel at is: zone remastering! AI could extremely quickly go through places like Deadwind Pass, Un’goro Crater, etc and modernize trees/ground clutter very effectively. Yes someone would still need to check and fix minor things it messed up on, but I feel like it would automate so much of the process and breathe back life into some of these old and wonderful zones many new players never get to experience. A lot of the leveling experience zips you through expansions at light speed; but when you get bored and feel like exploring, I feel like it’d be so good for the game to have it not feel like you’re traveling back to 2005 every time you leave the most recent expansion zones. It’d make it feel more like a whole world again.
Side note: I also think a potential use of AI would be plugging in the raw data and numbers and it doing the big tedious job of cross referencing all the assets to bring actual balance between the classes and maintaining it and tuning those damage numbers as new stuff and variables are added into the game.
I said this before a while back, but I think AI could solve a major issue that wow stumbled into. Transmog limits new playable races.
All* playable races after transmog was introduced have been “Rigged” onto existing 3D skeletons of other playable races. I think transmog is at the core of why we don’t have ‘cool’ new playable races with interesting rigs (like Sethralikk or Arakkoa). You’d have to pay someone to refit thousands if not tens of thousands of small 3d art assets onto these new rigs. Resize, reshape, move this spike a little to the left, etc. It would each a massive chunk of the art budget for the new race.
But AI, or some other machine learning, could take a pretty good pass at remodeling all the little bracers and boots to fit the new rig. It wouldn’t be perfect, but then you could pay a human artist to go through and QA the work and make small adjustments as needed.
All* isn’t technically true, with Evoker we got a new rig. But notice a few things:
they have very limited transmogable slots
the rig is very “human like” in that they stand up straight and have similarly human proportions
I think it would be interesting to have AI revamp an old leveling zone and then see what kind of flaws the players find. Like maybe some patches on the ground / trees aren’t physical, textures might be applied to the wrong objects, some geometry may inadvertently become traversable like the old way into Hyjal.
I asked chat gpt how to fix Druids, and it gave me patch notes that essentially increase our damage, rage, and threat gen by ~300% and returns mark of ursol.
I just want a silencing roar. Something like silence Sigils I can use to group up pulls.
AI isn’t a magic button. It requires you to train the model. So for Blizzard to apply that to their own game with their art style and own proprietary generative model would require an extensive project to get that functioning. This is not even including the legal battles behind this. They can’t just be running a public generative AI and use that in their games because they won’t be able to copyright them. It’s very messy.
Well in my honest opinion, I am surprised they havent been used in escort quests , as the NPC would do something unpredictable each time , rather than the same thing over and over.
Right, but they actually have training materials. You could use the existing rigs & transmogs to inform how to properly rig transmogs onto a model. Like showing it how a human moved this particular glove from human, to elf, to tauren, to troll, to goblin. Then ask the AI to guess how to apply it to Draenei and give it feedback compared to the actual artist’s work for Draenei. Given there are thousands of moggable models - this should be more than enough to train.
There are also NPC ‘transmogs’ as well that could be used to extrapolate more abstract models.
Also It’s not even generating new art, its deforming existing art assets to fit the rig. So it’s pretty narrowly contained.
I feel like AI hype is starting to outpace its proven accomplishments. It’s a generative tool. Blizzard devs already has generative tools in their toolkits. AI or no AI their toolkits will continue to evolve over the years.
(Er …not sure why I replied to you in particular, I must have had a different thought when I started my post)
I’m not saying it’s not possible, just that there’s the tech, legal and business layer that makes this a lot more complicated. I don’t think they can just use an existing API and build something on top of that. They’d need something in-house, and with that comes things like getting the talent for it, how they’ll maintain it and the cloud cost as well to run the models + storage etc…
I’m not “being trolly”. I think everyone is getting caught up in an “irrational exuberance” like we saw with ‘crypto’ and ‘NFT’.
I don’t doubt there could be tools that automate certain things, but ultimately I still want humans to pilot their development software. I want the creativity of humans in my video games, not whatever a piece of software generates.
You guys being so accepting of “AI” in everything is just going to encourage big business to invest in it even more, and fire humans because AI software doesn’t get paid, need a healthcare plan, benefits.
They’re going to go hard into automating as much as they can, and you all are going to cheer them for it. It will be a substandard product, and humans will lose their jobs.
Yeah agreed, they wouldn’t use out-of-the-box stuff that’s used to draw lude anime art. They would likely need to hire a 3rd party company that builds machine learning systems for exactly this type of work.
But outsourcing that work to a staff of 3-5 AI programmers from a 3rd party is orders of magnitude cheaper than hiring 25+ freshman artists to fiddle with every little boot model so it fits my ogre-toes. And that AI programmer staff is getting cheaper by the day.
I don’t disagree with you at all. I think a LOT of people and companies are “out of their minds” when it comes to what is and isn’t possible for AI to do.
But there are certain things that are just NOT being done. In wow we have this transmog example. It’s too cost prohibitive to hire humans to do this work - so it’s just NOT being done. So it’s not about firing a person and hiring a computer, its creating something that wouldn’t otherwise exist.
There’s really no creativity in making sure that your bracers start at the wrist and end at the elbow 125-times.
They could hire humans if they don’t currently have the manpower to do more. Don’t let the answer be “AI” can do it.
In the past few years they doubled the size of the wow team. They have money to hire people.
The industry just fired tens of thousands of people, for not much reason other than “it looks good to Wall Street”. If you encourage them on this AI path, they will fire as many people as they can get away with.
Right, but if they do that, they will have to pass that cost onto the consumer. So let’s say you’re the lead designer for the next expansion.
“Hey boss, we want to add a new Ogre playable race to the game”
“Ok Great!”
“So in order to allow them to transmog, we need to hire probably 25-50 new 3d modelers for 6 months, this means the new expansion will cost about $300 instead of the normal price”