Quote (NetflixAdaptationWidow @ Aug 23 2021 06:15pm)
I think that cars will likely be more efficient with AI during the "boring" parts of driving. Long stretches of uneventful highway. It will only become practical for city use after the vast majority of cars or even 100% of cars are also AI. The real challenge is that driving requires predicting what stupid humans will do and accounting for that. So you've not only got the issue of recognizing and modeling what's around you, but also the issue of having to model stupid human brains doing the same. We can do that because we have stupid human brains.
I think AI is going to be use-case specific, with general intelligence never really existing. So like, we will have an AI that assists doctors in reading charts to highlight risks, but probably not an AI that can adequately listen to patients on its own and key into difficult details, like when the patient is likely withholding information and how to account for that or probe for that information in a sensitive fashion. So you'll still have doctors interviewing patients and entering data, but the AI will help interpret the chart, test results, etc. etc.
I'm pretty sure that several car companies are already "better" than humans when it comes to the boring parts. They already get in fewer accidents per 100,000 miles driven than humans. You should read up on how Waymo trains their vehicles because they add some pretty crazy simulations (i.e. how to drive in a blizzard during an earthquake with Libyans chasing you with machine guns). Waymo has 20 million miles of driving experience on public roads and 20 billion miles in simulation.
I agree that general AI has a very long way to go. When I was in grad school, the holy grail for AI was computer vision and NLP. Since then, we've made some substantial strides and it's to the point where AI is competitive with humans with those types of tasks. Within 20 years, it would not surprise me if computers became better at those tasks.
As for the original topic, I think the real strides are going to be with manufacturing processes. I think in 50 years, we'll be able to create new organs on the fly, repair spinal injuries, etc. I don't think we'll be able to repair the brain though.