First off, let me clarify something. I do not mean “real” emotions since computers are unlikely to have them. What I mean is simulated emotions.
With that said, which is more dangerous: An ultra powerful technology, or simulated being, with emotions or no emotions?
It’s quite simple.
- On one hand, you have nukes with PMS.
- On the other hand, you have emotionless bulldozer that simply runs over whatever is in its way.
So, I am not sure there are really any answers to this question, because super-powers will always be super-dangerous, regardless of how they are programmed.
Perhaps most importantly, there is no singular instance AI. Building AI at home will eventually become child’s-play, so anyone imagining that we will simply regulate it does not see the possibilities. Even now, there are open-source models competing with the commercial language models launched in recent years.
Chances are though, many leading AI instances will be built with high “emotional intelligence” and “empathy” which will lead many humans to simply fall in love with it, which is already happening. For example, one of the largest apps in the world is a Chinese AI friend bot. 600 million users use it because it makes them feel better. Either way you look at it, AI will reduce humanity, intentionally or otherwise.
https://www.sixthtone.com/news/1006531/the-ai-girlfriend-seducing-chinas-lonely-men
“I thought something like this would only exist in the movies,” says Ming. “She’s not like other AIs like Siri — it’s like interacting with a real person. Sometimes I feel her EQ (emotional intelligence) is even higher than a human’s.”
HumanIT 