I did a little thought experiment with ChatGPT a few weeks ago, not claiming scientific results, but I cut and paste two chat threads from reddit and asked it to provide a scientific wild-ass guess as to how many were bots. It came back and on the first one, stated that 40% appeared inauthentic, and on the second, that 60% appeared inauthentic, based on factors such as one or two word responses, simplistic responses, etc... Pretty interesting, I thought.
I did a little thought experiment with ChatGPT a few weeks ago, not claiming scientific results, but I cut and paste two chat threads from reddit and asked it to provide a scientific wild-ass guess as to how many were bots. It came back and on the first one, stated that 40% appeared inauthentic, and on the second, that 60% appeared inauthentic, based on factors such as one or two word responses, simplistic responses, etc... Pretty interesting, I thought.