Raw LLM Responses

"ChatGPT can be bullied into saying whatever you want if you're persistent enough." Which means that on that topic, the problem started outside the GPT. Yes, there are specific instances of concern - but those are less in number and kind than what you get from joining online communities. Should we shut down reddit? There are entire communities devoted to pumping up each other's delusions. I think we should do our best to make AI as harmless as possible. But acting like it's got some great amount of harm relative to the internet as a whole is currently hogwash. Complaining about the spitballs flying from the sky and calling them evil yada yada when you're being shot with bullets.