Under Bavaria: Pepper taps a nonsense – Bavaria

“World premiere!” exulted Christian Nitsche, editor-in-chief of Bayerischer Rundfunk, on the talk show “Münchner Runde”. For the first time, a robot fed with artificial intelligence was allowed to join the discussion on the show. He went by the cool name Pepper.

Even during the rehearsal, the plastic little man talked very cleverly. It was almost touching when Pepper claimed to be a little nervous. Of course, that was a bluff, because the creature was certainly not nervous, rather it stared like a monkey when there was lightning.

When Nitsche asked if Pepper also spoke “a little Bavarian” (sic!), he hit a sore point, because Pepper answered confused. Presumably, the little man hasn’t been drummed into it that there is no such thing as a Bavarian language, only Bavarian, Franconian and Swabian dialects. Pepper was completely overwhelmed when Nitsche asked for a sentence “in Bavarian”. Now the robot wanted to show off, but it only spouted nonsense: “Of course,” Pepper crooned, “mia san die Bavarian and mia kenna gut mia.”

Oh dear, something went quite wrong with the programming. Bavarian language, Bavarian history, that’s where artificial intelligence fails. Mia san mia – that’s how the German champions in the Imperial and Royal Army used to sound, and today the German champions of FC Bayern cheerfully continue this custom.

Pepper’s brain consists of the chatbot GPT, which is currently causing a stir around the world. Besides Bavarian, he can do quite a lot. Within seconds he solves math problems, he writes homework and even love letters. People will soon be freed from the trouble of having to write their own texts.

Letting helpers like GPT do your homework is, of course, old hat. Grundhartinger Seppi, whose real name – to trample Pepper a bit – is slightly different here, relied on such bots 50 years ago. He let a Jopa plug jump when someone wrote his homework for him. There was no shortage of aspirants, and it often took a fight to decide who was allowed to take on Seppi’s essay. For the teachers, the deception was easy to see through, the various pig claws and the original variations in spelling were too revealing. Deception using chat GPT will certainly be more difficult to detect. Unless the essay is about “Mia san mia!”

source site