4 ms·
Eventually, artificial intelligence. Maybe not in the near future, but as technology progresses and every manufacturer strives to optimize their designs with t
by keiru 7y ago
Eventually, artificial intelligence.
Maybe not in the near future, but as technology progresses and every manufacturer strives to optimize their designs with the latest features, it will become an unsourmountable task to oversee every aspect of it (efficiently). I'm not talking about actively designing, but rather for warning/flagging for potential error. In very complex enterprises like global transport or building skyscrappers there is a lot to learn from experience and little human time, but it might be very cost-efective to train all-observing self-learning AI to look over everyone's shoulder, and warn you about using the right type of bolts, or how the coming heavy rains in Guatemala might affect your supply chain.
It's not that far-fetched when you realize it doesn't need to really understand anything, just be very good at playing word association and micromanaging.
- csours 7y agoAI is only as good as its training data and goals/success conditions.
- keiru 7y agoYes, AI is designed in very particular situations to fit ver specific tasks. I never meant there to be a single mind controlling the whole world. Today you could program an assisting AI that told you when "you missed a spot" when painting your house. It's simply not cost effective. But eventually driving and medical diagnosis AI, while imperfect, will have a better success rate that humans. Do you really think that won't apply to industrial production eventually, say in a hundred years?
- perl4ever 7y agoPreventing problems like this seems to me to be what AI as we know it is least suited to - failures of human organizations generally seem to be failures to choose the right context due to no individual grasping it, and my impression of AI today is that it doesn't even engage with the problem. You can have a computer program that recognizes cats, or plays Go, but nobody even thinks about "how do we make this same program spontaneously respond to a spilt glass of milk, or a hostage situation, or a fire alarm...", let alone take into account everything in the world while doing so. It feels to me like people have tunnel vision that is getting worse, and the "intelligent" software is inheriting that.