1/ This is about OpenGPT and every other language-model AI out there. #IOps #infosec #informationsecurity
Moral philosopher, professor emeritus of philosophy at Princeton University, Harry Frankfurt defines bullshit as follows:
Moral philosopher, professor emeritus of philosophy at Princeton University, Harry Frankfurt defines bullshit as follows:
2/ The one who LIES knows the truth, tries to hide the truth and tries to convince the listener. The one who talks bullshit doesn't care about the truth and only tries to convince the listener. The person or entity who talks bullshit, doesn't care about the truth at all,
3/ and to talk bullshit you don't even need to know if it is the truth or not. The only goal is to convince whoever is receiving the bullshit.
4/ Any answer that ChatGPT gives is complete bullshit. It may be right, but it doesn't know it's right, and it doesn't care if it's right. When using ChatGPT, or any language-model artificial intelligence system, it's good to remember what the overall goal of the AI is.
5/ Its aim is to CONVINCE YOU with its answer that it works. It is not looking for the truth, not giving good advice or being ethical. The main task of the language-model is to guess after every word it chooses, what would be the next word a real person would write there.
6/ It repeats that loop until some kind of answer is ready and then it produces that output to you. And that answer is complete bullshit in a philosophical sense, regardless if it was true or not. The system producing that answer has no concepts of "truth", "responsibility" or
7/ "justice". It churns out as much believable bullshit as possible, it gets more believable by the hour.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
