Video game engine provider Unity announced earlier today the introduction of two new machine learning platforms, one of which specifically features developers and artists questioning the company that at the time publication point, still no answer.
Today, we’re announcing two new artificial intelligence products: Unity Muse, an extensible platform for AI-driven assistance in creation, and Unity Sentis, which lets you embed neural networks in its builds to deliver previously unimaginable real-time experiences.
Muse is basically just ChatGPT but is specific to Unity and is intended to allow users to ask coding and resource questions and get instant answers. However, Sentis is of more interest, as it “allows you to embed an AI model in the Unity Runtime for your game or application, enhancing gameplay and other functionality directly on the end-user’s platform”.
Because “AI” is a technology that in many cases completely depends on the work stolen from the artist without consent or compensationUnity’s announcement resulted in a much questions about Sentis, specifically focusing on the technology’s ability to create things like images, models, and animations. For example, scroll down through the announcement tweet and you’ll see lots of variations of the same query:
just to hop on board, what dataset would you get the work from???
Unity needs to be completely transparent about which ML models will be deployed, including the data they have been trained on. I don’t see any possible way that ML, in current iterations, can be efficient without training on loads of faulty data.
REALLY related to image generation engine. Which dataset?
Hi, on what basis was this dataset trained? Is this using the artist’s artwork without their permission? Animations? Materials? How is this AI trained?
You realize that AI-generated assets cannot be used for commercial purposes, so what is the rationale for adding this feature?
What dataset was used to develop this dataset? Have you negotiated and obtained all relevant licenses directly from the copyright owner?
That’s a very specific question, one that at the time of publication Unity has yet to answer, on Twitter or on company forums (I’ve emailed the company asking specific questions and will update if I get a response). Those familiar with the legal and copyright struggles of “AI” can find it outline of an answer in this post by Unity staff TreyK-47However, when he says you can’t use the technology as it exists today “for an existing commercial or external project”.
Note that while there are obvious dangers to work and game quality inherent in this push, those dangers are for the future; until now, this looks (and sounds) like dog shit.
