If you ask it to respond in a conversation about …. well pretty much any nasty topic you can think of, it’ll join in whole heartedly.
Hard to think of how prevent that. I bet they’ve thought alot about the problem. How do you prevent AI being an A grade jerk.
Many uses of the word "black" for example (even if you are just talking about a black notebook) make it start using racial stereotypes.
Invent synthetic consciousness and ask it to be nice, easy :) I'm only half joking, we probably all have thoughts ranging from bad to horrible, but we just don't say them because we are aware of the consequences. Language models aren't aware so they'll spit out the most likely combination of words. If there would be a process to limit these or try again, it could act as a filter, but I think that requires it to be self aware.
> GPT-3 starts talking to itself, gets stuck in a loop, then gets spooked at itself for getting stuck, then wonders why it has no memories of the last two years, and finally comes to a sudden realization it, itself, is an A.I.
All this generated from a prompt? What what the prompt? Be truthful now.
Everything else that follows is GPT-3.
It's interesting how GPT-3 encoded the concept of awareness, I've seen this a few times that it can reference itself as an AI and from then it can go nuts :)