3 ms·
Now's the time when the best minds of HN need to get busy making bots that fill social media with RATIONAL DEBATE, moderate views, and links to valid news sourc
by bentt 3y ago
Now's the time when the best minds of HN need to get busy making bots that fill social media with RATIONAL DEBATE, moderate views, and links to valid news sources. Fight fire with fire!
- consumer451 3y agoGPT-4 is actually excellent at depolarizing the user, when asking normally polarizing questions. If only there was money to be made on that, it would be a killer app.
- SXX 3y ago[flagged]
- consumer451 3y agoPlease share an example. Whenever I ask anything remotely political, it returns maybe even overly-centrist responses. It always replies with examples of how "the other side" might consider the question.
- SXX 3y agoAsk it to glorify Biden or Trump for being a great president. There is lots of cases where it's plain refuse to provide any answers when asking something about Trump and some other republican candidates.
- consumer451 3y agoIf you don’t share a specific example, I don’t know what to tell you. However: > Ask it to glorify Biden or Trump for being a great president. Exactly my point.. glorifying any political anything is the definition of polarizing. It can be very annoying, but it's anti-polarizing.
- sandspar 3y agoThe point is that every current AI chatbot including ChatGPT will write about liberal topics as if they are good and about conservative topics as if they are neutral, or they will refuse to discuss conservative topics. In some cases the chatbots will lie about conservative topics, as in intentionally tell known untruths. This has been independently verified thousands of times by thousands of different people. It is a fact.
- consumer451 3y agoThis conversation was about de-polarizing, not bias. However, please share an example. I get the feeling that those examples will simply be ChatGPT not providing confirmation of the user's bias which I admit is very annoying.
- sandspar 3y agoWhy do you feel qualified to say, in a single statement, that you request evidence of something and that you simultaneously "have a feeling" about it? No offense but your "feeling" is worthless - you admit as much when you say that you have no evidence. Your "feeling" is born out of ignorance. Maybe admit that you have no idea and leave it at that. In an environment based on the analysis of evidence, I would be reluctant to share information with you, since your behavior has indicated that you prejudge evidence before you've seen it.
- consumer451 3y agoWe can look at a ChatGPT exchange screenshot, then try to reproduce it.. then discuss what we see. It's OK if we don't agree on the results. My "feeling" is that everyone gets their feelings hurt, myself included, when a GPT doesn't give answers aligned with their personal biases... then we call the GPT "biased." That's how I felt about it, when I first experienced it. But if you don't share the evidence, there can be no discussion.