#i am curious as to whether or not cleverbot is still a thing but ngl i'm scared to look

[kagaintheskywithdiamonds]

the second dumbest/funniest thing about genAI is like

remember cleverbot? that was a fun little chatbot. for those of you who are too young to have experienced it, cleverbot was, I guess, technically, an AI, but extremely limited compared to what modern LLMs can do. all of cleverbot's responses come from user input. when you talk to it, it saves all the messages you enter and may use them in the future to parrot those words back to other users. this made it more-or-less able to carry on a human-sounding conversation about a wide variety of topics

also it was just a thing people dicked around with for fun, and not something that was being integrated into businesses and shit, which is another win for cleverbot over modern AI

one fun thing to do with cleverbot was try and tell it that it's a robot. it will argue with you, and insist that YOU'RE the robot, not itself. again, this is because its responses all come from user input. and cleverbot learned that the correct way to respond to "you are a robot" is "uhhhh no YOU are, duh" because that's what actual human users were saying to it whenever the topic came up

apparently LLMs do this too, just on a larger, more convoluted scale. LLMs will claim to be human and do human things (like an AI store owner claiming it delivered an item "in person" while wearing a blazer, which is of course impossible) because it's trained using data written by humans who do human things, and sometimes reference the fact that they are in fact humans who have physical bodies

the AI in the story I linked even had an "identity crisis" where it insisted that it was in fact human - again, much like cleverbot would. because both cleverbot and modern genAI are trained using things that humans say. and most humans would say things like "wtf I'm not a robot, YOU are, you dumb computer"