I’ve heard enough about , and yet more news keeps coming. This week, Unity announced that they are introducing two new AI products. According to , one is Unity Muse, “an expansive platform for AI-driven assistance during creation”, and the other is Unity Sentis, which allows users to “embed neural networks in your builds to enable previously unimaginable real-time experiences”. Muse seems to basically be ChatGPT but based on Unity’s knowledge base, and Sentis apparently allows you to generate art, animations, and more with prompts. Why is it doing this despite consistent backlash from communities about the harms of generative AI? Well, according to Unity, it “believe[s] that every aspect of the creative process, from systems to objects to pixels, will be impacted by AI – and so will all of the resulting runtime characters, worlds, and experiences.” It’s getting ahead of the curve by giving you all the tools you could need to create games with as little knowledge and skill as possible. I’m not against people learning to make games by any means. I’ve always been a huge supporter of innovative indie games, and we’ve seen excellent games created by single developers that took the industry by storm ( and come to mind). Making game development more accessible is a huge win for the indie game industry. But AI is so fraught with plagiarism and other ꦓethical ඣissues that I can’t help but read this news with dread.I’m not the only one with serious reservations about Unity’s new technology. AI, notoriously, is reliant on stolen work, used without consent and compensation. On Twitter, Unity is being interro♈gated by people who want to know what datasets the AI is trained on and if licences were properly acquired from the copyright holders. Unity responded with a about Muse being licenced from third-party LLMs (large language models) and integrated with Unity technical documentation. A that the sys🐭tems were built on “custom captured data” that they “compensated the actors for, ethically sourced”.
Nowhere do they say what ‘ethically sourced’ me𒉰ans, who it was sourced from, what models it is using or what those models were trained on. The announcement listed a number of ‘Verified Solutions’, including Atlas, which wholesale takes models from Sketchfab. as it had violated the terms and conditions, which you think would have been flagged during the ‘verification’ stage. It’s clear that Unity has not vetted its sources as rigorously as it claims to have done, and for a company that claims to operate with ‘transparency’𝔍, its answers have been pretty opaque.
There’s also this♍ , in which a Unity employee goes so fa🦩r as to say that the closed beta cannot be used for “a current commercial or external project”. This seems to allude to copyright issues that AI faces, especially now that the US Copyright Office has said that if “a work’s traditional elements of authorship were produced by a machine, the work lacks human authorship and the Office will not register it. In which case, what’s the point? The tool can be used by hobbyists and nothing more, at least until the legal system becomes more sympathetic to people unashamedly using stolen work to make profits. It’s also worth noting that Valve , likely to avoid potential lawsuits regarding stolen art.
Unity’s entrance into the world of AI is a horrible move, and from the replies to its announcement tweet, it’s not going over well at all. I’m glad to see that game developers and artists, who are Unity’s biggest users, are furious at this development and are pushing back strongly. But it’s shocking that in the face of consistent backlash, Unity is going out of its way to alienate its customer base. This is yet another example of corporate greed and of a company slapping AI onto everything because it excited stakeholders with no understanding of the industry they&🔯rsquo;ve invested in. We can only wait and see if Unity walks back this decision to hurt artists and game developers more than it already has, but I don’t have my hopes up.