The Finals and AI

It's always cool to see how the world manages to fit in one final story about the buzzword of the year into the headlines before everything is said and done. For no new release can be free of it's own injection of controversy in order to 'spice things up'- so to speak. In this case the game in question is the hugely popular destructibility championing first person shooter- The Finals, which finally released on all platforms in the middle of the game awards after wooing over the shooter public with it's solid preview events and striking gameplay focuses. And not being much of a multiplayer gamer and having not actually had the time to try out The Finals for myself, I feel comfortable enough to confidently say that the game is total meandering trash- I'm joking, of course. People seem to like it enough so I'm sure there must be something to all the hype. But becoming the budding Internet darling is not enough, it would seem, to shield The Finals fully from all the drama it's stumbled into.

So as the name might suggest 'The Finals' presents itself as something of a 'fake game show' tournament murder game, presumably broadcast over some sort of dystopian society who's remnant cities are the arenas that the fighters blow to pieces. A common premise, you've seen it before. Fitting in with this theme the game features a couple of announcers who talk over the action, announce shifts in the game, provide slight colour commentary, the usual guff out of team based shooter games. These announcers in particular sound a little off, as though the vocal director did not take the time to coach them on each line. Or the actors are only semi-fluent in English and stumble on inflections still. Or... or almost as though their dialogue is rendered entirely through an AI. But that couldn't be- wait, it totally is- isn't it? That's what we're currently embroiled within this blog, isn't it?

That's right! It didn't take too long for the accusations to be made and the most scary thing is that unless you're listening out for it- it's easy to fool your ear into thinking the performances just aren't very good. Pay some attention however, and spend enough time around other AI voice videos, and you hear the taletale giveaways. The lifeless delivery, missing obvious inflection points, the rushed pace of certain lines. There's no human behind it and- it actually hurts the performance. But only ever so slightly. Given time, and more data, it's pretty likely we'll get models that aren't noticeable in the future, and maybe even play through entire games without knowing there were talking computers programmed in until the credits thanking ElevenLabs. And you can picture the kind of havoc that knowledge is currently wrecking on the voice actor community.

You see, Voice Acting is a pretty tight knit community of actors who aren't actually the richest people in the world, struggling for each and every job, usually between bouts of unemployment, trying to make enough of a name for themselves to be self sufficient. And, well, lets just say that the prominent use of AI in a large video game is kind of like walking on the grave of acting as a profession, it sends chills across everyone's spine. A voice actor's voice is their most precious and sought after commodity, it is their valuable instrument, to counterfeit that is to introduce fake bills into a bank vault- it taints the entire batch. Not my best analogy but do you even need one to understand how worrying people with a genuine stake in the continuation of human voice actors finds the AI of 'The Finals'. Although to be fair to everyone, the game was in early states of development for a long time whilst being shown off to the public- so in that vein I guess it's acceptable for placeholder text to be AI work, before the final performances with real people are brought into the fold, right? Right? They're not going to hire real actors, are they?

No, as it so happens the Finals actually consider the AI voices part of their vision for what the game should represent, and what they want to convey. As in, they would like to convey how much quicker and cheaper it is to churn out endless lines as if on a factory conveyer belt- ain't that just sweet of them? Obviously, there's a large question of ethical sourcing here. There's so talk around claiming that the team did actually hire actors to provide the base voice performances that were then fed into an AI machine, but were they compensated accordingly for use of their talents in perpetuity? Almost certainly not. And that is the sticking point. Voice Actors aren't just worried that their talents are going to go to waste, or that they're going to have their voices stolen (although both are very valid worries to have) they also are concerned about the prospect of getting screwed out of living wages. 

And believe it or not, this isn't the only such issue right now involving AI. Or rather, believed to be involving AI. The recent entry in the Naruto Ninja Storm series has featured baffling vocal performances so staggeringly off-putting that people seem to have come to the conclusion that the English cast has had their performances replaced with AI composites. The English voice actors of the famous characters themselves, when reached out to, appear baffled as to the performances and seem pretty adamant that they never delivered lines quite so bizarrely, and considering the status of the game as a union project this could, should the claims be valid, be battlegrounds for a AI versus VA war in the coming near future. Or at the very least this could be the last straw before an industry wide strike is deployed.

AI is already something of a snake eating it's own tale for most of it's current uses. AI only really works to an effective rate then it's being fed data of real people, and the more AI starts to proliferate into our own data the more it feeds into bad output which feeds into a cognitive model collapse. But when we extend that to the way that AI voice synthesis works, I'm not sure we can breathe such a sign of relief. It's hard to say how ElevenLabs and it's copycats and competitors train their models, but we can probably assume that unless they are exceptionally lazy it wouldn't be hard to prevent fake AI generated speaking examples to taint their training pool. Which means that AI is only going to become more convincing at mimicking human speeches and soon you won't be able to pick which is which from a Twitter compilation.

All and all there really isn't anything profound to pick out of this story other than- maybe we should have tighter reigns on the use of AI. I understand there's huge potential for revolutionising the way that games are made with the help of AI, but we've had generative technology for decades now and still the best that a AAA company like Bethesda can do with it is Starfield. Maybe tools like these shouldn't be waved haphazardly about when people's livelihoods are at stake, and maybe the various rigors that would otherwise be brushed by with AI are necessary friction points wherein the shine of a polished vision is finalised. It's just a shame to have a great up-and-coming game be tarnished by such avoidable nonsense. I'm talking about 'The Finals', apparently the Naruto game is just a mess. Pay your workers, not your machines.