A A strange piece of software has recently arrived on the PC gaming store Steam. And “software” seems like the cleanest way to describe it. A full-blown life sim, existing somewhere between a science project and a haunted fish tank of sorts, nLife: Motion-Learning Life Evolution If it were not an unusual factor it would probably disappear without much impact. Several years ago some of its creators were completely roasted on camera by one of the real giants of Japanese animation.
In 2016, Hayao Miyazaki, director of films like Princess Mononoke and Spirited Away, was shown new technology that used AI to animate models. Encountered by a zombie that slammed its skull on the ground to move its head and twitched its body like a fish, Miyazaki declared that what he saw was “an insult to life”. It’s hard not to watch the clip without feeling a little hurt — but now, a decade later, the ashen-faced developers in that room have recovered enough to make their work widely available.
Judging by the conversation around launch, at least some people downloading EnLife are doing so in the hope that it might provide some clues about the current state of video games’ relationship with AI. Putting aside the often widespread use of that term, it’s certainly something worth trying to understand, whether it’s due to job losses caused by AI or blamed on AI, or the sheer number of games created with the aid of AI models now landing on storefronts like Steam.
However, there is a problem here. And the thing is, EnLife itself is such a delightfully inconsequential thing that it’s hard to read much of anything into it.
NLife promises players an evolution simulator where “AI-powered block creatures move in unpredictable ways.” For the most part this involves placing various creatures in a small environment and then watching as they learn to move around.
Visually EnLife is pure Frutiger Aero, offering landscapes of green valleys and sparkling waters, which can be soothing images MRI technicians sometimes encourage you to look at during long scans. Sonically it’s equally unimpressive: with a series of bloops, blips and popping sounds, we’re thrown straight into the soundtrack of a million 00 day spas.
This desire to calm down extends even to the level of mechanics. During a free morning with EnLife you can place a variety of simple organisms in an environment and then feed them food that will encourage them to reproduce or mutate. You can expand your territory and then attract creatures to the water or to the air to create more variety. There are plenty of things to unlock (including a shadow tech tree that’s got you covered if you’d rather destroy your digital sea monkeys than watch them flourish) but the ecosystem is kept simple. It’s a game of seeing how things crawl towards food.
The thrill of it all probably comes down to the “how” in that last sentence. This is where the game possibly includes a slightly mysterious use of AI. And it’s true that, after a few hours of play, you will have NLife’s strange little creatures discovering new types of joints and body arrangements as they swim, fly and generally wander around looking for food.
(AI has a history of using creatures to move around, incidentally, often using small neural networks and computer developments. In 2009, UK-based game technology company NaturalMotion developed a project In which a bipedal model learned to walk using advanced neural networks. The company was later purchased by Zynga in 2014.
There are two problems, however, the first being that the focus on unlocking the skill tree gives the early hours of NLife a mindless clicker game feel that it struggles to overcome. The second thing I gather that people study procedural generation is sometimes called the “porridge problem”.
The porridge problem, first formulated by author, developer and academic Kate Compton, relies on the fact that every single bowl of porridge in the universe is unique. Just not in a very interesting way. Similarly, when non-living organisms discover a new way of rolling, jumping, or flapping their bodies toward food, they are still moving toward food. This makes for a game that either really pays attention to small changes in detail, or completely zones out and just enjoys the floaty aesthetic. During my time with EnLife, I’ve typically started with the first approach and then discovered after 10 minutes that I’d switched to the second approach.
The more I played EnLife, the more I started thinking about something I heard from one of the first AI researchers I spoke to in 2013. The real value of AI, he explained – and I’m paraphrasing here – is that it could one day become an entirely alien form of intelligence. This means that we humans will have a different kind of mind lens through which to view ourselves and our quirks and cognitive fallacies – things that can be difficult to recognize when you’re talking to people who think in a similar way to you.
Since then the apparent focus of many companies like OpenAI appears to be the complete opposite: creating exemplary plagiarism machines that sometimes exist only to tell people what they already want to hear.
It feels like EnLife is filling a very small and specific niche. As far as I’m aware, it’s not trying to hide its AI use. It’s really trying to draw attention to how it uses AI to allow its little creatures to move around. In other words, it’s not just a game born from AI. This is a game that is actually about AI to some extent. It wants to be a little alien. It does not want to hide itself or be ungrateful; It seeks to investigate AI’s innate capacity for otherness.
