Virtual reality playground and chat service VRChat has decided not to implement an AI filter within its virtual metaverse.
Earlier this year, VRChat released a survey to players of the online game asking about multiple aspects of the virtual reality experience’s future. Within the survey, players were asked about AI filters which would allow people to filter out or knowingly avoid AI-generated content.
In response to fan feedback, the developers of the game have decided not to follow through with the wants of its players. In an official blog post, VRChat confirmed that “95% of survey respondents want some kind of label for AI usage”. The company confirmed that 21,595 players filled out the survey, so around 20,515 asked to block out GenAI.
The developers argued that survey “figures are weighted towards VRChat’s player base”… which, well, of course they would be. Additionally, the company claimed that not all users would honestly label their AI-generated content which would cause other non-AI creations to be attacked by accusations of AI use.
“While we think some creators would honestly label their content, we think others would try to pass off generative or assisted work as completely authored by hand,” the company said. “Likewise, we could see this creating ‘sleuthing’ behavior, where folks would attempt to sniff out ‘AI’ content or might harass other creators, even if they are genuinely not using any AI tools in their workflow. We think this potentially creates negative behavior spirals we don’t want to encourage.”
The studio claimed that one of the biggest reasons behind the decision to not have an AI filter was because few people could agree on how they wanted it to work. “Wanting a label isn’t the same as avoiding everything that would carry one, though. Only a third (33%) would avoid an ‘AI-assisted’ world on purpose. And 37% of everyone who wrote to us said some version of ‘AI as a tool is fine, as long as a person is doing the creative work.’ Further, a small number of respondents want disclosure so that they can learn what tools they could explore on their own.”
“You don’t all agree, though,” the developer continued. “Some of you want AI kept out of VRChat entirely. Others use it as part of how they build, for making tools, for translating UI and text to languages they don’t know, or to write code when they don’t know how. Many fall somewhere in between.”
Furthermore, VRChat is updating its harassment rules to include harassment over the use of AI tools. Under the new rules, players who flame other people for generating work instead of creating it will be suspended or even banned as a result.
While the studio is thinking about adding a “human-authored mark” to promote human-made content, they are refusing to add an AI filter. While many users have specifically asked for it, the studio believes that “the definition of AI is quickly evolving and nobody is fully agreed on where this line is”.
“We’re not adding them,” they said. “Filters get set once and forgotten, and then you never see that part of VRChat again. In addition, a filter assumes that labels are correct, so it inherits all the problems we talked about above. This all results in world discovery and community issues that we’re not keen on having.”
