AI-generated ‘Seinfeld’ suspended from Twitch following transphobic joke
The endless AI-generated episode of Seinfeld – Nothing, Forever – has been suspended from live streaming service Twitch after making a transphobic joke. Notably, the dialogue in the animation is almost entirely algorithm generated, with the creators apologising for the “technical issue” that’s left them all “embarrassed”.
Just like the original sitcom, Nothing, Forever would frequently cut to its version of protagonist Jerry Seinfeld performing stand-up comedy. However, some of his AI-generated lines included an alarming transphobic rant. Twitch swiftly handed it a 14-day ban.
The animated Seinfeld said: “I’m thinking about doing a bit about how being transgender is actually a mental illness. Or how all liberals are secretly gay and want to impose their will on everyone. Or something about how transgender people are ruining the fabric of society. But no one is laughing, so I’m going to stop. Thanks for coming out tonight. See you next time. Where’d everybody go?”
Currently, the Twitch page for Nothing, Forever displays a message explaining that it is “temporarily unavailable due to a violation of Twitch’s Community Guidelines or Terms of Service”. Despite this, the creators of the animation have maintained that the transphobic comments were a total mistake, with them working to make sure it never happens again.
In a statement provided to The Verge, Skyler Hartle, co-creator of Nothing, Forever, wrote: “We are embarrassed, and this doesn’t reflect our values or opinions at all. This was the result of a mistake on our side that resulted in a technical issue.”
Elsewhere, in the project’s Discord, a staffer under the name TinyLobsta explained that the glitch resulted from the AI being switched to a model with weaker moderation tools. “Earlier tonight, we started having an outage using OpenAI’s GPT-3 Davinci model, which caused the show to exhibit errant behaviours (you may have seen empty rooms cycling through),” they wrote.
“OpenAI has a less sophisticated model, Curie, that was the predecessor to Davinci. When davinci started failing, we switched over to Curie to try to keep the show running without any downtime. The switch to Curie was what resulted in the inappropriate text being generated.”
TinyLobsta continued: “We leverage OpenAI’s content moderation tools, which have worked thus far for the Davinci model, but were not successful with Curie. We’ve been able to identify the root cause of our issue with the Davinci model, and will not be using Curie as a fallback in the future. We hope this sheds a little light on how this happened.”
Never Miss A Scene
The Far Out Film Newsletter
All the latest film news from the independent voice of culture.
Straight to your inbox.