4 ms·
Gemini models won't touch a lot of things that are remotely political in nature. One time I tried to use GPT-4o to verify some claims I read on the internet an
by resters 2y ago
Gemini models won't touch a lot of things that are remotely political in nature.
One time I tried to use GPT-4o to verify some claims I read on the internet and it was very outspoken about issues relating to alleged election fraud, to the point where it really got in the way.
I generally find it unhelpful whaen models produce boilerplate meant to couch the response in any way.
- petesergeant 2y agoIn what way was it outspoken? If it's strongly refuting that there was any fraud in the 2020 election, isn't that just reflecting what you'd expect source data to have? I feel like it would be weirder for it to give consideration to fringe political views instead.
- resters 2y agoIt could be, but it feels more like the model would respond, but is then stopped from doing so by another model whose job is to censor it.
- Jensson 2y agoThe problem is when a model returns a premade answer that the developers put there, those aren't very convincing or good arguments. It would be better if it returned a dynamically generated response adapted to your question, so this isn't about sides but neutering the models ability to respond well. Its like when a human stops engaging and thinking, and instead just returns a tribal message, at that point the human isn't worth talking to, the same thing goes for these models. You don't want them to return tribal copy pasted messages like that.
- petesergeant 2y ago> The problem is when a model returns a premade answer that the developers put there Sure, and do you have any examples of that happening with Western models?