Up in the enormous halls of the northern half of Koelnmesse, the crowds are still being wowed by glitzy stands and demos of the latest games, not least the long-awaited Hollow Knight: Silksong.
But in the southern half, the business-only section of the show is drawing to a close. And having spent the past four days dashing between appointments with CEOs and developers, there is one sentiment that has remained consistent among almost everyone I spoke to.
We need to make games quicker.
It’s refreshing to hear. After months and months of gloom and panic across the industry, as layoff announcements arrive as regularly as bad-news buses, it feels as if everyone has finally centred on a plan.
Shorter development times will of course mean lower costs
It’s a simple one. Rather than spending half a decade or more working endlessly on one title, the idea is to instead make games in one or two years, maybe three at max. And if they’re not quite polished enough for a full release by then, they can be popped into early access instead.
By far the biggest expense when making games is salaries, so shorter development times will of course mean lower costs – in theory. And that means not betting the farm on every single release.
If a game that’s been in development for two years fails to land at launch, it’s still a big blow. But it’s nothing like the existential crisis of launching a flop that’s been in the works for five, six, seven years.
There’s the advantage, too, that quickly made games can be adapted to suit current trends, avoiding the pain of, say, launching a live-service shooter years after the genre has been saturated.
Of course, it’s one thing to say you want to make games more quickly, and quite another to actually do it. More to the point, how do you do it?
One option is to make games that look worse. Given how super-detailed graphics seem to be far less important to a younger generation raised on Roblox and Minecraft, this would seem like a fair enough strategy.
Why bother spending days, weeks, or even months modelling super-realistic satsumas when your audience would be satisfied with a crude orange daub?
Yet there seemed to be little appetite for this strategy among the people I spoke to at Gamescom. Perhaps it’s an unwillingness to fly in the face of conventional wisdom in an industry where frame rates are often fetishised. Perhaps it’s more about simple pride in the craft.
So what’s the alternative? One option is to use AI to speed up the development process. And it’s an option that more and more studios are taking up.
AI is the games industry’s dirty little open secret – the majority of people I spoke to said they were using AI in some form or another.
Very few were employing AI to generate finished assets for a game, the kind that gets you that shameful little ‘AI Content’ label on Steam. But many were using it at some point in the development process.
AI is the games industry’s dirty little open secret
Utilising AI to generate snippets of code was a popular choice. In addition, a fair few people are using AI to generate concept art early in the process, letting them quickly iterate ideas.
Everyone was adamant that AI should be used as a helper tool, rather than as a replacement for human skills.
Some people were quite open about the use of AI in their games. Others were far more coy, going rigid when the dreaded word came up, as if worried their secret might come out.
They have reason to be afraid. The outrage caused by a snippet of AI-generated text being found in The Alters – along with the more serious problem of poorly AI-localised text – is one example of why developers are wary of talking openly about AI.
Yet the fact is that AI is already in widespread use across the games industry – and it seems absurd for developers to live in fear forever. What’s needed is an open discussion of how AI should be best used. What’s needed are agreed best-practice guidelines.
For example, should AI-generated art be off-limits in finished games? Or is it fine as long as the data set is trained on assets wholly owned by the studio? These are the kinds of questions that need to be discussed.
The next few years will entail a process of collectively deciding how to proceed. But love it or hate it, it’s quite clear that AI isn’t going away any time soon.
Whether AI actually enables games to be made more quickly, however, remains to be seen. I have my doubts – the temptation with effort-saving technology like this is always to do more, rather than do it quicker.
Maybe the goal of making games faster will take a while to achieve, and might well require a change in thinking. But at least everyone has agreed on a plan.