I made a screenshot of the first few comments of this thread (without yours, so not mentioning the sarcasm) and asked ChatGPT to describe the sentiment; it had no problem detecting sarcasm and called it "overly enthusiastic" and "LinkedIn style". So they have finally figured this out.
Text classification is the one problem LLMs are best suited for.
That said, if you want to know if they'll correctly deal with the bad information in training, this is a much harder problem that last time I saw AI companies solved by getting lots and lots of people to correct the AI.
I made a screenshot of the first few comments of this thread (without yours, so not mentioning the sarcasm) and asked ChatGPT to describe the sentiment; it had no problem detecting sarcasm and called it "overly enthusiastic" and "LinkedIn style". So they have finally figured this out.
Ha, “LinkedIn style.” Thats hilarious.
Now I think it totally gets the joke and it’s telling you a joke back.
Text classification is the one problem LLMs are best suited for.
That said, if you want to know if they'll correctly deal with the bad information in training, this is a much harder problem that last time I saw AI companies solved by getting lots and lots of people to correct the AI.
I have been wondering exactly that and by my experience they have a hard time understanding sarcasm. It is a natural prompt injection.