If these cars really work, it’s a mighty big thing. I don’t think it would be an exaggeration to describe it as the biggest breakthrough in computer technology since the very invention of computers. (One cynical question I’m fond of asking is whether there has really been any great progress in software technology in the last 40-50 years, and this is the first thing I’ve ever seen that might give a decisive positive answer to it.)
Also, a tantalizing question is what the economic (and other social) consequences will be once technologies like these start proliferating and replacing human workers across the board. Before this recent talk about self-driving cars, I had believed that such technologies would run against nearly-AI-complete problems and thus remain unavailable for a long time to come, but assuming this isn’t just empty self-promotion, we might be looking at some very radical (and potentially very nasty) developments quite soon.
Admittedly this is a subjective judgment, but I really think it’s incomparably greater. Once you take into account the steady improvement in computer speed and storage capacity, scalable and ultimately global computer networking is a straightforward development at some point along the road. After all, networking protocols consist of precise algorithms that are complete and self-contained in and of themselves, which just need powerful enough machines and network links to be executed. (In fact, the complexity of the present protocols comes only from cumulative clever optimizations for performance.) In contrast, these driverless cars are supposedly able to grapple with messy real-world situations and make decisions that can’t be reduced to state machines in any obvious way. I can’t think of any comparable technology that works well in practice.
If these cars really work, it’s a mighty big thing. I don’t think it would be an exaggeration to describe it as the biggest breakthrough in computer technology since the very invention of computers. (One cynical question I’m fond of asking is whether there has really been any great progress in software technology in the last 40-50 years, and this is the first thing I’ve ever seen that might give a decisive positive answer to it.)
Also, a tantalizing question is what the economic (and other social) consequences will be once technologies like these start proliferating and replacing human workers across the board. Before this recent talk about self-driving cars, I had believed that such technologies would run against nearly-AI-complete problems and thus remain unavailable for a long time to come, but assuming this isn’t just empty self-promotion, we might be looking at some very radical (and potentially very nasty) developments quite soon.
Bigger than the internet? Cars could already be driven, but people can’t do internet stuff without the internet.
Admittedly this is a subjective judgment, but I really think it’s incomparably greater. Once you take into account the steady improvement in computer speed and storage capacity, scalable and ultimately global computer networking is a straightforward development at some point along the road. After all, networking protocols consist of precise algorithms that are complete and self-contained in and of themselves, which just need powerful enough machines and network links to be executed. (In fact, the complexity of the present protocols comes only from cumulative clever optimizations for performance.) In contrast, these driverless cars are supposedly able to grapple with messy real-world situations and make decisions that can’t be reduced to state machines in any obvious way. I can’t think of any comparable technology that works well in practice.