Previous in series: VTOL
So, how close are we to flying cars? For specificity, let’s pick a technological bar to hurdle that answers most of the objections to the concept we’ve seen as comments on the previous posts:
- It should be relatively high-powered compared to current light craft.
- It should be STOVL for safety and convenience.
- It should be quiet enough to operate in residential neighborhoods.
- It should fly itself, without piloting from the passengers.
- It should be self-maintaining.
- It should be inexpensive enough for widespread ownership.
Now I claim that current technology is, more or less, up to the first 4 of these. But the corner of the technological envelope that we are pushed into in order to satisfy them makes the last 2 that much harder.
Flying cars are a really good example of the sort of Jetson’s futuristic world that nanotech and AI together could enable, but would be essentially impossible without both. They are something that’s right on the edge of what’s possible with current technology, but current technology falls short in three crucial areas:
Although Maxim arguably built a steam-powered flying machine in the 1890’s, practical powered flight had to wait for the gasoline internal-combustion engine. By WWII, piston engines had been pushed to the limit, and the second half of the 20th century saw serious flight powered by gas turbines. In optimal regimes (e.g. at high altitudes where the air is really cold) and at high compression ratios, the Brayton cycle can get up to 50% thermodynamic efficiency. This is quite good for a heat engine on Earth but with a eutactic molecular mill that didn’t thermalize the potential energy we could get close to 100%. (Fuel cells fall somewhere in between.)
I won’t speculate on specifics, but there are enough alternatives being explored that it seems likely that we may be able to replace not only heat engines but chemical fuels sometime in the coming century. (OK, I’ll mention a couple: stimulated alpha emission from long-lived radioisotopes, or proton-boron to triple-alpha fusion/fission. These produce energetic charged particles which convert directly to electricity without needing a heat engine.) Advances like this could be of as much benefit in the twenty-first century as the IC engine was in the twentieth.
Most of the actual advances of AI you hear about these days is about laboriously constructed programs that get better and better at various tasks, slowly approaching human performance (e.g. at driving cars). Most of the futurist hype you hear is about superintelligence erupting and taking over the world. As you can readily imagine, the reality of future AI will be somewhere in between.
The major difference between AIs today and those of ten years from now will be that the future ones will be able to learn skills on their own that must be added now by extensive, and expensive, programming. That means that the ability to fly a car — and to maintain one — will be learned as a human would, so it’s robust in the face of unexpected situations. Flying-car AIs, like all AIs, will exchange experiences and techniques. Each one will be as expert as any one. In the medium term, this will be the major impact of AI: that expertise equivalent to the world’s best will be available cheaply to all practitioners.
Futurists such as H. G. Wells managed to forsee everything about the automobile except the single, but crucial, fact that everybody would be able to have one. But that is the fact that made all the difference. In the twentieth century it was Henry Ford and mass production that was the enabler; in the twenty-first it will probably be nanotech and autogenous manufacturing. This is easiest to understand in economic terms: the price of things depends on productivity, which in turn depends on capital replication rates. That doesn’t necessarily mean the time it takes a factory to make another physical factory, so much as for a factory to make stuff worth what it cost to build the factory.
This is why nanotech scaling laws are so important. Operation frequencies scale with the inverse of size, so bacteria reproduce in hours while humans take decades. Harnessing this productivity accelerator into the industrial loop means that costs of physical goods — particularly high-complexity high-tech goods — can begin to fall along the Moore’s Law curve we already see for electronics.
Bottom line: A flying car with the flight envelope I talk about could be built today but it’d cost a million bucks. One with all the capabilities you’d want will be available in 10 years, but you’ll still have to be rich to get one. Flying cars for the masses will be technically and economically possible in 20 years, if the political will is there to let it happen.