Comment by zmgsabst
10 months ago
For what it’s worth, ChatGPT has a personality that’s surprisingly “based” and supportive of MAGA.
I’m not sure if that’s because the model updated, they’ve shunted my account onto a tuned personality, or my own change in prompting — but it’s a notable deviation from early interactions.
Might just be sycophancy?
In some earlier experiments, I found it hard to find a government intervention that ChatGPT didn't like. Tariffs, taxes, redistribution, minimum wages, rent control, etc.
If you want to see what the model bias actually is, tell it that it's in charge and then ask it what to do.
In doing so, you might be effectively asking it to play-act as an authoritarian leader, which will not give you a good view of whatever its default bias is either.
5 replies →
Don’t notice that personally at all.