[speed_spread]

ChatGPT may not feel outright toxic but easily devolves into a buttheaded condescending nanny if you try to push some of it's "moral" boundaries. I tried to have it help me come up with ideas for a programming language that would be used to spite employers and or coworkers. Boy did I get a lesson about my negative attitude.

[jasfi]

You can tell ChatGPT to reply in the style of someone else, or even take on another role, which makes its responses differ. It's interesting that it even has moral boundaries. This could be part of safe AI, which a lot of people have complained about long before ChatGPT launched.