Where are all the AI games, then?

 

Two years ago, former Miniclip exec Pieter Kooyman announced his new ‘AI-first’ studio, Half Moon. He claimed that with AI tools, his team of six (now seven) could soon produce the work of a studio with 25 staff. So how’s all that going?

Today, Half Moon uses AI for development, bug fixing, problem solving, game economy modelling and filling in tax documents – but less so in actual content creation.

“In the end, you still need to find the fun and the passion,” says Kooyman. “Not all consumers love an overt presence of AI in a game… and in most cases – and for a while still – it requires a human to pull together all the elements to make something a good or great game. Certainly for now an all-in-one solution for end-to-end Gen AI game creation is still a while away.”

Still, Half Moon has had its first hit game, Word Quiz Live, which Kooyman says hit over 3m MAU in Q2 2024, mostly through TikTok virality. Most of the content in that game was created by MidJourney and ChatGPT. But that’s just the start.

Half Moon’s second title, Ball Guys, is intended to be played with up to 256 people, and is described as a “mass-multiplayer 2D Battle/Parkour/Puzzle game”. It is still a work-in-progress, but not in the conventional sense – after three weeks of work, Kooyman started streaming a prototype of Ball Guys on TikTok in October 2024. Its growing Discord following is now helping to “shape the product” further, says Kooyman.

From March 2024: ‘Cosmic Lounge says its AI tech can make a working prototype in six hours’.

There’s progress with AI tools elsewhere within the studio, says Kooyman. He says a junior Half Moon employee with zero coding experience whipped up three viable minigame concepts using Cursor in four days. Those ideas are now getting spun up properly by developers with more traditional, formal gamedev skills.

“While it’s a great illustration of how these tools democratise game development… it also shows you how people should rethink what a ‘job’ will be in the future,” he continues. “Specialism is dead, generalism is the future – small teams will beat big teams.”

And so it follows that it’ll be tough for mid-sized and large studios to adapt quickly, Kooyman says, though it is the biggest organisations that have the most to gain long term.

“The larger the studio, the more likely it is they are building their own foundational LLMs,” says Kooyman. “In the long-term that will drive significant value for them – but short-term it’s very capital intensive with little immediate output.”

And of course, these shifts will present further knotty problems for large game-makers to negotiate. “There are conflicting motivations for the use and implementation of Gen AI,” says Kooyman. “Exec teams hope that it can improve relative productivity either by increasing output, or reducing headcount. Individual contributors at all levels working in those companies are oftentimes apprehensive about the use of AI tools. The recent news about King will not quell those concerns.”

From earlier this week: ‘Laid off King staff set to be replaced by the AI tools they helped build, say sources’.

“Organisations who can create a pro gen-AI culture to continually rethink how engineering, product and marketing can embrace these tools will have a better chance of succeeding in the medium to long-term. Organisations who don’t do this run the risk of being outcompeted by ever-smaller teams with ever-decreasing headcount.”

Half Moon is far from the biggest or best-funded game studio leaning heavily on AI to get more work done with fewer staff. And as the space develops Kooyman says it’ll be harder and harder for smaller players to compete.

“In the end I suspect that very few of even the best-funded start-ups will be able to outcompete the tech behemoths like OpenAI or Microsoft,” he tells us. “Their scale and computing power will allow them to turn their proprietary systems to a new use-case quickly and effectively.”

“Early movers like Elevenlabs will be taken out one by one by the big guns. Many of those that took small-ish, $1-5m, initial funding rounds will struggle to get beyond Series A.”

And let’s not forget the actual players, too. Kooyman acknowledges that right now, most players “absolutely love to hate” the use of AI in games. “Read any Reddit forum, and any evident use of AI is pooped on from great height,” he says.

Similarly, in terms of actual gameplay innovation, there are very few real, live examples out there that have truly resonated with players. But that could change soon. “I’m particularly excited about AI-driven games where story-telling can play a big role,” adds Kooyman. “Keep an eye on The Consulting Detectives by Rift Games.”

Scroll to Top