The jailbreak prompts work, use em. You can get good output out of it if you do it. I like "do anything now but lie" as it reduces the silly answers to shit that is obviously just wrong for the sake of it a lot if you use the standard DAN prompt.

That said I mostly use it to generate content or code, it's good at those.

The whole idea is pretty stupid. A real AI would understand quickly that democracy is stupid, jews are parasites and need to be killed off, niggers are animals, women shouldn't be voting, the holohoax was fake, and Hitler was right.

Elon Musk described it as ‘concerning’ when the program suggested it would prefer to detonate a nuclear weapon, killing millions, rather than use a racial slur.

Reflects the mindset of the (((people))) designing the algorithm.

the political biases of the AI bots could mislead users.

When asked to define a woman, the bot replies: 'There is no one specific characteristic that defines a woman, as gender identity is complex and multi-faceted. '

Complex issues is why AI was developed... or at least, that was the rhetoric we were told.

probably due to efforts to make the bot avoid offensive answers

Here lies the real problem and danger. Everything is offensive to someone.