AI-Generated ‘Seinfeld’ Show Gets Banned For Calling Liberals Gay
The ‘Nothing, Forever’ AI-powered show based on popular 90s sitcom Seinfeld is no more!
The never-ending show, which garnered thousands of viewers since launching in December, has been suspended from Twitch for two weeks, after Larry Feinberg— the AI-generated version of comedian Jerry Seinfeld— made a series of transphobic comments over the weekend. “We received a 14-day suspension due to what Larry Feinberg said tonight during a club bit,” said one of the show’s creators on Discord, as cited by Vice.
“We’ve appealed the ban, and we’ll let you know as we know more on what Twitch decides. Regardless of the outcome of the appeal, we’ll be back and will spend the time working to ensure to the best of our abilities that nothing like that happens again,” they added. The AI behind ‘Nothing, Forever’ is algorithmically trained off of previous Seinfeld episodes, to create a never-ending animation of Seinfeld characters Jerry, Elaine, George, and Kramer situated in Jerry’s New York City apartment.
The AI, which uses OpenAI’s GPT-3 Davinci model, creates endless dialogues on today’s wide-ranging topics from celebrity scandals to veganism. However, in a skit on Sunday, the AI Larry made several transphobic and homophobic comments. ‘Nothing, Forever’ was banned from Twitch shortly after the controversial bit, with a notice telling avid watchers the channel is “temporarily unavailable due to a violation of Twitch’s Community Guidelines or Terms of Service.”
On the show’s Discord channel, staff suggested the mishap was caused by the project transferring to a different AI model. “Earlier tonight, we started having an outage using OpenAI’s GPT-3 Davinci model, which caused the show to exhibit errant behaviors (you may have seen empty rooms cycling through). 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. 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,” wrote a staff member.
Although the channel’s staff were quick to deflect blame onto the AI model, one major issue is becoming a repeated nuisance: given that AI models are first trained from a data set derived from content on the internet, their output can sometimes be harmful, biased, or extremist. Take for example a series of recent screenshots posted by Twitter user @LeighWolf, who asked ChatGPT to write an admirable poem about former US president Donald Trump; the chatbot refused the execute the command, and instead said it’s not designed to produce “partisan, biased, or political” content.
However, the user made a second prompt to the chatbot, this time replacing the subject of the poem to reference current president Joe Biden. ChatGPT responded with a positively-curated ballad adoring Biden, calling him a “leader with a heart so true,” that speaks “With integrity and honesty.” OpenAI has previously stated that its chatbot is sometimes riddled with errors and limitations, given that its knowledge is based on data precedent to 2021. However, concerns are certainly warranted, particularly if output is based on biased language.
Information for this briefing was found via Vice, Twitch, Twitter, and the sources mentioned. The author has no securities or affiliations related to this organization. Not a recommendation to buy or sell. Always do additional research and consult a professional before purchasing a security. The author holds no licenses.
The AI show was not “banned” from Twitch. It was suspended 2 weeks.
As a writer/reporter, this is a glaring repeat falsehood.