Seinfeld on Twitch proves AI shouldn’t write comedy

After viewers tagged the segment with transphobic remarks, Twitch banned it. Nothing, forever for two weeks. Hartle, embarrassed and apologetic, said he would be more careful in filtering topics for offensive ones. But he didn’t promise that Larry would get more funny stuff.

Nanyung Peng, Associate Professor of Computer Science at the University of California, Los Angeles, studies the ability of AI to be creative and is a co-author of articles “A single structure for creating puns with the principles of humor” And “Surprise pun generation“. She says that the unfunnyness of AI is due to the fact that it uses a probabilistic model to determine the most expected idea, and the humor is based on unexpected answers.

This seems like a bad excuse. I’ve worked in sitcom writers’ rooms, and a lot of the jokes are just math. Theoretically, for this you need to be able to use machine learning. In fact, Peng once taught an AI model the rules of humor, as comedy writers have tried to explain. Such theories are usually the rule of three and the inconsistency theory. “Our machine was able to generate ‘the greyhound stopped to get a haircut,'” she told me. I felt bad for not laughing, which seemed to wear her out. “It’s not that funny. But this is a few. When we saw this result, we were excited.”

The task, according to Pan, is enormous. “I don’t think people really understand jokes. There are no theories where you can use them and then you will become a stand-up comedian. Part of it is really talent,” she says.

Comedian Whitney Cummings, who had a robot made in her likeness for a 2019 Netflix special, wasn’t surprised that the artificial intelligence was telling terrible jokes. Why are people shocked that robots aren’t funny? Most people are not funny. The only funny robots are Roomba when they get stuck under the sofa,” she says.

Cummings is generally in favor of robots; she even keeps a robotic version of herself in her house. But she doesn’t expect it to make her laugh. “Comedy is one of the few things that is so characteristic of the human being,” she says. “A comedy about trauma that comes from a person’s life experiences and how they deal with it. Robots cannot be injured.”

When I asked Spike Feresten, who wrote for Seinfeld from 1996 to 1998 why did he think Nothing, forever wasn’t funny, he suggested asking the AI ​​why it wasn’t funny. But when I logged into ChatGPT it said it was not available because it was full. Strangely, however, on the left side of the page, its glitch was explained in the form of an AI chat suggested by “Write a stand-up comedy about ChatGPT status”. The closest thing to a joke in it was:

Comedian: “I think I’ll have to talk to my cat for now. At least he doesn’t have a queue.” (laughs and claps)

When I sent this to Feresten, he replied, “It’s like asking why Spock isn’t funny.”

In fact, when I was later able to log into ChatGPT and asked why it wasn’t funny, the bot basically said the same thing as Feresten, only with less humor: “Although AI can recognize patterns and generate responses based on them, He doesn’t have a sense of humor like humans do. He doesn’t feel emotions, doesn’t understand context, and doesn’t pick up language nuances the way humans do.”

However, he was able to sum up this article in a few seconds: I should have added more jokes.