The Limited Times

Now you can see non-English news...

Fake Seinfeld copy banned for anti


The »Nothing, Forever« series was actually supposed to continue without interruption with the help of artificial intelligence. But a joke against minorities brought the project to a temporary end.

Enlarge image

A scene at a comedy club interrupted the project

Photo: Screenshot Twitch / Watchmeforever

It is a development that has often been seen: if artificial intelligence is fed material from the Internet, it does not take long before they also make racist or otherwise unacceptable statements.

This fate has now befallen a project that had received a lot of attention on the Internet: »Nothing, Forever« is an algorithm-controlled joke machine that imitates the successful and cult series »Seinfeld« of the 1990s and has attracted a small fan base in recent weeks attracted.

But anyone who accesses the "Watchmeforever" channel on the Twitch streaming platform currently only gets an error message that the channel is not available due to a "violation of the community guidelines".

As the US magazine "Vice" reports, a scene in a comedy club is the reason for the ban.

There, the fictional main character Larry Feinberg begins to ponder that transsexuals endanger the cohesion of society.

The character also claims that leftists are gay and want to impose their will on society.

Two weeks break

This didn't go over well even in the entirely fictional and automated series universe: "Nobody laughs and that's why I'm going to stop now," said the virtual comedian before the automated series switched to another scene.

But apparently viewers had already taken offense at the performance and reported the channel to the Twitch streaming platform.

They then routinely switched the channel offline for two weeks.

The series makers want to try to end the ban beforehand with an appeal, but definitely want to continue as soon as the channel is unlocked again.

Actually, the company OpenAI had quite successfully trained such lapses in their GPT-3 language model, which the project uses.

Nevertheless, there is now a community of people who want to use all sorts of tricks to get artificial intelligence to emit offensive or even dangerous content.

In the case of the Seinfeld parody, however, this was not necessary.

Instead, it fell victim to the popularity of ChatGPT.

Since millions of users are now experimenting with the chatbot again and again, consuming enormous amounts of computing time, access to the GPT-3 AI running on the same servers has been very unreliable in recent days.

Since the endless series is dependent on a constant supply of jokes, the makers quickly switched to another language model that was less mature.

After all, the series makers from the »Mismatch Media« collective believe they have found a technical solution that should prevent such gaffes in the future.


Source: spiegel

All tech articles on 2023-02-07

You may like

Trends 24h


© Communities 2019 - Privacy

The information on this site is from external sources that are not under our control.
The inclusion of any links does not necessarily imply a recommendation or endorse the views expressed within them.