Game Development Won’t Be Revolutionized by Generative AI Just Yet. A VIDEO game requires a lot of tedious, difficult effort. How is that possible? Since developers create entire worlds, it makes sense that the gaming industry would be thrilled with generative AI. A tiny crew might create a map the size of San Andreas by using computers to handle the tedious tasks. Crunch disappears, and games are released in a complete state. A new era is coming.
Game Development Won’t Be Revolutionized by Generative AI Just Yet
This story has, at the very least, two interconnected issues. First, there is the logic of the hype itself, which, whether intentionally or unintentionally, tends to view automating artists’ labour as a form of advancement and is reminiscent of the frenetic gold rush over crypto/Web3/the metaverse.
Second, there is the disconnect between these assertions and actuality. The venture capital firm Andreessen Horowitz published a lengthy analysis on their website in November, when DALL-E seemed to be everywhere, praising a “generative AI revolution in games” that would decrease development times and alter the genres of games being produced.
In the month that followed, Andreessen partner Jonathan Lai wrote on Twitter about a “Cyberpunk where much of the world/text was generated, enabling devs to shift from asset production to higher-order tasks like storytelling and innovation,” and he proposed that AI could make game development “good + fast + affordable.” Lai eventually received so many angry comments on his mentions that he had to create a second thread to admit that “there are obviously tonnes of issues to be handled.”
Patrick Mills, the acting franchise content strategy director at CD Projekt Red, the company behind Cyberpunk 2077, says, “I have seen some, frankly, absurd statements about things that is purportedly right around the corner.” “I saw comments claiming that AI could create Night City, for instance. I believe that is still some time away.
Even many who support generative AI in video games believe that much of the hype around machine learning in the sector is out of control. Julian Togelius, codirector of the NYU Game Innovation Lab and author of numerous publications on the subject, calls it “crazy.” Sometimes it seems like the worst kinds of crypto bros came over here and said, “Generative AI: Start the hype machine,” as the crypto ship was sinking.
Togelius clarifies that generative AI can and should be used in game creation. People are overestimating what it might be able to do, which is the problem. Sure, AI could create some generic weaponry or dialogue, but level design is difficult compared to word or image generation.
Generators that produce a face with crooked ears or a few lines of random text are OK. However, no matter how fantastic it appears, a broken game level is meaningless. He describes it as “bullshit,” adding that it must be manually fixed or thrown away.
Basically, no one wants level generators that function less than 100% of the time—and Togelius has discussed this with a number of developers. They ruin entire titles by making games unusable. That’s why it’s challenging to simply integrate generative AI, which is challenging to govern, he argues.
Read More:
- ChatGPT Is Causing Universities To Reconsider Plagiarism
- Ex-Twitter Employees Are Puzzled By Elon Musk’s Discontinued Laptops
