So Asimov’s Laws were basically the first system prompts? Chatbots get ‘don’t be harmful,’ robots got ‘don’t kill humans.’
When the output is physical instead of text, alignment matters a lot more.
"Most ancient accounts of the reality beyond our everyday experience were not only false, they had a radically different character from modern ones: they were anthropocentric. That is to say,
at the physical hub of the universe as well. Those two kinds of anthropocentrism – explanatory and geometrical – made each other more plausible, and, as a result, pre-Enlightenment thinking was more anthropocentric than we can readily imagine nowadays."
The tension with AI isn't new. It's safety vs. opportunity. We faced this with nuclear power too.
When safety wins absolutely, progress loses absolutely.
The real question isn't if AI is risky, it is, but if our fear will cost us good things we can't even picture yet.
Ads in cars
Left - Ford's publication US20240289844A1
Right - Jeep's implementation
Next update: Your steering wheel will be locked until you watch this 30-second ad about car insurance
@PaytmMoney Can you guys solve ticket number 7015126507? Playing with your customer support from 2 months, with their out of world solutions like "try again", " reinstall app"
Maybe ask Deepseek on how to solve problems