Summary
Xboxis the latest companyto get involved with generative AI, launching a new partnership with InWorld aimed at aiding development. People can often overreact to AI as an ominous bogeyman, quick to forget thatAI has been part of video games since the start- it’s how the ghosts chase you inPac-Man, or why the other paddle moves in Pong. But there are real reasons to be scared of Xbox’s latest move, and when it arrives, we shouldn’t fall into the trap of mockery.
To be clear, I’m not out here to defend AI. I’m not saying we should give it a chance and see where it goes. But we’ve been very complacent on AI so far, and our usual disaffected memery won’t be enough to overcome these challenges.
The key word inMicrosoft’s statementis “generative”, meaning it will be using AI to create things, which is when you get into murky waters. InWorld and Xbox plan to “build AI game dialogue and narrative tools". How the software will do this is by “turning prompts into detailed scripts, dialogue trees, quests, and more”.
Here’s the thing - the first example of this being used will stink. It will be janky and stilted, with flat delivery, awkward phrasing, and dead-eyed stares at the camera. And we’ll all laugh at it. We’ll forget that that description could easily go toStarfield, or that of the two mainGame of the Year contenders(Tears of the KingdomandBaldur’s Gate 3), one can’t produce round edges or remember where you left anything, and the other frequently stutters and takes minutes at a time for the computer to realise it’s its turn. All games are a miracle, right? So we forgive these things. But how will we react to AI?
We wrote this two months ago when Geoff Keighley was roundly criticised for promoting AI tools. Many people discovered Xbox’s partnership through a separate tweet from Keighley, this time through the official The Game Awards account.
Again, this is not to defend AI. This is not to implore you to give it a chance, or to let AI plagiarise whatever it needs to before it can generate something worthwhile. But our first reaction is going to be to mock it for being bad. Then it will get good. And then we have nowhere to go.
We saw this with AI art. ‘Ha, it looks so stupid, look this one has seven fingers!’ only worked for a short while.Now it has the right number of fingers and it looks fine. Some AI art, taken individually, is even good. It looks like the thing it’s meant to look like. But they all have those hyper-saturated shadows, the plastic uncanniness, the colour blur. They all become dull clones of themselves, and thus become boring.
Here are some examples of the great works of art AI has created
They all look like the same thing over and over and over because AI can’t really ‘create’, it can only ‘generate’. Two synonyms with very different truths. But it looks fine, and since everyone on the ground floor was focussed on how stupid the fingers looked, any serious criticism is now playing catch up as all the general public is an art form that has rapidly improved to the point of usability.
Games are significantly harder to make than a still image, but I still expect a sharp learning curve. AI generated art takes the path of least resistance and steals from everything around it, so it makes early progress quickly and breaks ground never. It’s a fast learner that hits the Peter Principle as soon as it’s vaguely usable. But if all we do is talk about how crummy the early iterations are, our protests will quickly be aged out as the AI learns.
The ‘Peter Principle’ is the idea that people are promoted to the level at which they become incompentent, and then remain stuck. AI art is good enough to look like a copy of a picture, but it’s unlikely to ever do much else.
The first AI generated things Xbox (or the numerous other studios rolling out AI measures) will be terrible. And we’ll all have a good laugh about them - but soon they’ll get better, and if all we did was laugh, the joke will be on us. Another obvious argument is that these things cost jobs, but cruel as it may sound, that’s also unlikely to land with the general population. Just this week we sawDestiny fans celebrating the Bungie layoffs, proffering the argument that the devs have ruined the game.
We have seen a dangerous separation of art and artist in gaming, where development is a thankless job and all the positives in a game exist by magic and all the negatives are the results of lazy and woke devs inserting their own agendas rather than making the game fun. AI has no agenda, these gaming molluscs will say, and therefore costing jobs (and thus keeping development costs down) will be a price many gamers will happily pay.
It has to come back to the difference between creating and generating. There are some experiences in gaming that require a human touch. The characters we most deeply connect with are not born of algorithms or demographic focus groups, but from a writer putting a part of themselves or their lives down on the page. Gaming moves forward when devs try things nobody else has done before - a task AI is incapable of as all it does is borrow and replicate.
Performance is an even harder aspect of human creation for AI to replicate. If you wanted to be particularly cynical you could look at our current media landscape whereAndor, aStar Warsshow about a ragtag jailbreak, is dismissed as highbrow and pretentious and consider that perhaps people would be satisfied with warmed over, generic AI writing. I don’t think it’s true, but I could see where you were coming from.
Few, however, would look at those strange videos of theKendall Jenner/MrBeast AI chatbots, orlisten to the esports AI commentators in The Finals, and think that getting rid of human actors would be anything less than a huge downgrade. Baldur’s Gate 3’s charismatic cast isa key reason why it is held in such high esteem. It’s hard to imagine whatever plopped out of ‘AI prompt: sexy, vampire, British’ beingas beloved as Neil Newbon’s take on Astarion.
The soul of gaming is a harder point to make than ‘haha silly fingers’ or the cold hard numbers of job losses. But AI art can draw fingers now and the fact is gamers don’t care about layoffs. But humans have the advantage in art. AI might be faster and cheaper, but humans are better. Humans can create new things and explore deep in their hearts. Maybe some games can be made by a computer doing things faster and cheaper. But they’ll be empty. The games we care about, the games that stick around, need new thought, they need heart.
It’s easy to mock early AI attempts at replicating human art. It’s fun too. But it’s not very effective, especially with AI’s rapid learning curve. It’s clear that gaming is going to continue to explore AI to find a way to make it stick, and it won’t be scared off as easily as it was withNFTs- AI is bad in a lot of ways butit’s not the scam NFTs were, it can create ‘value’ in ways pictures of monkeys or owning your guns in Rainbow 6 cannot.
To protect the industry against it, we need strong messaging that AI cannot do what a human can do. Not that itshouldnot, or that it looks goofy when it tries, but that deep in the soul, itcannotdo it. Remember that when Xbox andeveryoneelserolls out their AI monstrosities with too many fingers.
