Let’s think about slowing down AI

lesswrong.com

2 pointspaulpauper3 comments

"If you fear that someone will build a machine that will seize control of the world and annihilate humanity..."

We don't understand how this may happen. It's science fiction at this point. It makes no sense to slow down when we don't understand the problem. We need to do more research until this becomes clear, then consider controls

The other possibility is just that a state or private actor that didn't slow down will be first.

People always want to fixate on something popular. As you say, there is literally no pathway right now from AI to annihilating humanity, but we like do dwell on in.

Meanwhile, disease research, antibiotic resistance, actual global conflict or proxy conflict between nuclear superpowers, climate change (whether the direct effect or misguided attempts to geo engineer it away), etc all have (maybe at different levels of probability) at least a clear causal way they could seriously harm people. But we'd rather talk about how a chatbot will somehow kill us.

Reply on news.ycombinator.com