I saw that one. Makes sense to me. If you're relying on an AI bot to answer your customers' questions, the answers should be considered binding. This is going to lead to some more fun cases...

LOL AI doing the right thing, give people their money back, Air Canada thought they're slick not using people to handle queries, its cheaper now look, pretty sure theres going to be tonnes of this stuff, no one checks under the hood or even knows how these things work but pushing it in their businesses

Air Canada is a net negatif on the canadian people. They get freshly new printed money every time they get in trouble, since 2020 that government owns a big stake in that bankrupt company so they are one of those "too big to fail" zombies out there. And canadians wonder why their tickets, rents, fuel and grocery bills are high. The money is being printed to feed those clown zombies

Wonder if it would've gone down the same if the chatbot wasn't GPT-powered but gave a wrong response.

I am sure we will see many more cases like this in the future.

Dude. I wonder if we're on the cusp of a whole era that has people who just hang out all day and try to get free services from chatbots. I can even imagine companies hiring "chatbot hackers" to bring down competitors. ...not saying that would be ethical, but it's probable and also pretty funny.