Generative AI in game development, and ‘automation anxiety’

Photo by Alexandre Debiève on Unsplash

Dr. Aleena Chia from Goldsmiths University in London lectures on cultures of creativity in game production. At the recent King’s Festival of Artificial Intelligence conference she spoke about how AI will feature in future game production, and how game workers and leaders can cope with “automation anxiety”.

Here are a few edited excerpts from her presentation on how the creative community are talking about, and dealing with, the notion of automation in game development.

“Creativity was once viewed as future-proof, but recently we’ve seen a lot of discourse around the idea that, no, it’s not future proof, and that causes anxiety, a kind of primal, pit of your stomach pre-sentiment about AI eating up jobs.”

“We see studies, like the one from Goldman Sachs, predicting that as many as 300 million full time jobs globally could potentially be in jeopardy. According to the public discourse, It’s taken as a given that it’s already happening to us, and that we have to compensate for that.”

“William Gibson said the future is already here. It’s just not evenly distributed. And the same thing goes [for AI production]. As we build the automated future – and we are building it – there are winners and losers. It is not evenly distributed. There are stratifications, especially in games.”

“There are different disciplines in games – an artist’s work and a programmer’s work and a business person’s work are different from one another. There are different kinds of creativity depending on your tasks. Studies rarely account for this kind of granularity, and the differences between, for example, whether you’re a technical artist, or you are an asset artist.”

“A thing that people don’t really talk about is that creative jobs have always been insecure. They’re always patched together from different kinds of contract work, freelance work, multitasking, in the games industry and beyond.”

Uniquely human

“On the other hand, we hear a lot about creativity being something that’s uniquely human – humans have a soul, a self, free will. So that’s why we can never be replaced in terms of creativity. It’s so ingrained, but we should stop saying that, right now. We do see humans not being replaced, but not for those reasons, per se, but because humans are more adaptable. They are more flexible [than computers]. They have perceptual ways of seeing the world that are not easily computable.”

“Technical artists talk about how, even after you generate AI models – say, a rock hitting a wall, made by a tool that is designed for that job – you always have to go back in and tweak. But the human in the loop is not the soul or the consciousness. The human in the loop is here to respond and adapt flexibly to things that a machine can’t do. There are many examples of this throughout the generative pipeline.

“This is what is known as automation’s last mile. It’s always easier for actual human beings to come in and lift that robot vacuum cleaner up the stairs to clean the second floor. Likewise, there are always glitches in automated processes. And that will always generate new tasks for humans.”

Creative definitions

“Creativity is always political. If you ask an artist and an engineer to define creativity, you’ll get very different answers, and that’s fine. But if one definition comes to dominate all the others, then there is an imbalance.”

“The drive to disempower workers is always there – from industrialization to automation. They exist to drive efficiency, but also to reduce the power that workers have over the process. Automation takes investment, but it always seems to make sense for managers to achieve more control and more efficiency.”

“But efficiency is not always the goal. Independent game developers don’t talk much about efficiency, but about creativity expression, working with other people, making cool things. Game development is a multidisciplinary, interdisciplinary enterprise that involves lots of skill-sets coming together. It’s not just a technical solution to be solved.”

“Toolmakers must automate with purpose. They will define the kind of value of different kinds of creative work. What counts as shoveling assets into pipelines? What counts is something that can just be automated away. Is it [generating] hair and baseboards? Or is it narrative? The granularity of control is important on the level of how people work, but also how people are able to control their own jobs. In many ways this is not really a technical problem at all, but a managerial problem – the recognition that the creative process is stratified.”

“What can we do to reduce the risks, reduce the harm of automation? Harm reduction comes at the end of a process. But why not interrogate the process right now and make sure that it is safe for use before it is deployed out in the world. Automation is not some force that’s out there in the world, that’s washing over us and that we can’t do anything about “

You can see the full presentation here.