My last post about Google got an enormous response. A lot of people focused more on the driverless cars than I meant to (I was just using that as one example of AI monetizing) but it is an interesting topic. Presumably it’s why I got so many comments (hundreds between my website, email, and other places). I’ll give my thoughts on some of them.
First, some people said “that isn’t interesting, science fiction writers have been talking about this for decades.” Maybe they have. The closest thing to my vision that I’ve seen is Minority Report. I remember a scene with cars zipping by at incredible speeds. I think they were levitated though. I haven’t seen anything that resembles what appears to be the near future.
It may have been written about, in fact it probably has many times, but like 99.9% of the population I haven’t read it. I’d be happy to if someone has a good recommendation. Though I’m always optimistic about the genre I find most sci-fi to be poorly written crap. Like most niche genres (fantasy, western, romance, etc.) the fact that it caters to a pre-disposed segment of the population (in this case, people who like to think about science and technology, of which I am one) means the writing has to be much less good to achieve a level of success.
In my list of benefits of driverless cars, I compacted the timeline greatly. I see the transition occurring in the following three stages.
Stage 1: All cars on the roads have drivers.
Stage 2: Some cars have drivers, some don’t.
Stage 3: All cars on roads are driverless.
The line between stages 1 and 2 will be very blurry, but it will be drawn in the history books somewhere around now. (Humorously, many people said “cars can’t drive around on the streets by themselves yet,” apparently oblivious to the fact that Google’s cars already are.) As some commenters mentioned, we’re moving increasingly in that direction now. Dynamic Cruise Control matches the speed of your car to that of the car ahead of you. Cars can parallel park themselves. They can hit the breaks to prevent you from getting in an accident when the driver ahead of you stops abruptly, and they can warn you when you’re coming out of your lane. These aren’t the techs in labs either; they’re in models you can currently buy, many of which aren’t very expensive.
The line between stages 2 and 3 will simply be government regulation. Stage 2 will last for years, and maybe even a few decades, and then it will end when the government says “within 10 years all cars on public roads must be driverless”. The real shame is that there’s no politically and technologically feasible way to skip a protracted Stage 2. All the best stuff (like the highly increased average speed) will have to wait until Stage 3, and to be honest Stage 2 is far harder of a technological problem. If we were able to start from the ground up, designing both our system of roads and our automobiles we could easily have driverless cars now.
A lot of people pointed out that government regulation will inhibit Stage 2. I’m not so sure. For one, it’s clearly legal now or Google wouldn’t be doing it. I can’t imagine they’d be dumb enough to have cars driving themselves around the streets of San Francisco without asking their legal team what the risks were in advance.
Also, this technology must be in the late stages. Google doesn’t want the PR nightmare that would ensue if one of their cars hit a pedestrian or t-boned some old lady at an intersection. If Google has these cars out there logging hundreds of thousands of miles, you can be sure they’re pretty damn safe.
And really, why wouldn’t they be? We’ve had driving simulations for decades. Once the car’s sensors do a good enough job of modeling the reality around them, the rest is nothing more than a video game. If an AI can race competitively in the latest Gran Turismo why can’t it drive around New York? Is it really that hard to detect other cars, pedestrians, stop signs, traffic lights, etc?
There’s a strong chance that by the time regulation even addresses this issue it will be solved. The NTSA won’t have to guess about safety, they’ll have hard data. My guess is it wouldn’t be hard for computers to do a better job driving our cars than we do. It turns out we’re pretty awful at it, and the proliferation of smart phones is making us materially worse. Texting and driving is believed by many to be both more dangerous and more common than driving drunk.
Some people said we’ll see full automation in airlines first, for regulatory reasons. I don’t believe it. People are decidedly and predictably irrational when it comes to air travel. Certainly we could have this technology there, but one airline crashing would cause a groundswell of opposition that you’d need 10,000 cars crashing to accomplish.
Take 9/11 for instance. After the planes hit the towers, air travel decreased markedly, and miles driven shot up. More people died as a result of the extra driving due to the terrorist attack than on the planes during it (at least according to one researcher). People fear flying in a way that is psychologically understandable but statistically irrational.
On the opposite end of the coin, people hate driving. In fact a long commute actually makes you more unhappy than a bad marriage. A number of people like driving and there will certainly be racetracks for them, but I think most would agree that the benefits of complete automation vastly outweigh the hobby of a minority.
A lot of people didn’t like my thought that urbanization will reverse. I’m using “urbanization” to mean specifically living in cities and not suburbs. People will mention that urbanization has increased, but in America that is only because suburbs are typically counted. In fact in the industrialized world people have been leaving cities for surrounding areas for decades. Now in the US, the population in the suburbs of the 20 major cities exceeds that of the population in those cities. I did not mean to suggest we’d all go back to an agrarian society, I simply see the world trending toward one giant suburb.
People gave a number of reasons why urbanization would continue instead, but almost all of them are based upon notions of proximity that become antiquated in a Stage 3 world. Proponents of urbanization pointed out that population density leads to variety. This is true now, but that’s an artifact of geography. The real benefit to density is proximity (almost every other characteristic is negative) and the only benefit to proximity is travel time. That travel doesn’t have to be by car, and perhaps even despite cars being effortless and flawless in terms of safety some people will still want to walk for health reasons. (Humorously enough, not having to drive will, I think, lead to people having something like an exercise bike in their car to work out on while travelling.)
Take restaurants for instance. An Ethiopian restaurant can be found in any big city like New York or Chicago. Good luck finding one in a small city like Akron, where I live. The reason is entirely travel times. When choosing a restaurant, the time required to get there is a deal-breaker. I may be in the mood for tapas, but I’m not ever going to drive two hours to eat it. I’d have to be really in the mood (and the restaurant really good) to even drive half an hour.
This puts a practical limit on the restaurant’s customer base. To start an Ethiopian restaurant, you need to have enough people who like Ethiopian food (or will try it and then come to like it) who can get there within some amount of time, let’s say 20 minutes. In a densely populated city like New York or Chicago that exists. In a sparsely populated one like Akron that doesn’t. Triple the speed at which everyone drives, and you’ve now tripled the geographic customer pool. People could spread out to 1/3 of the current population density and still have the same amount of variety they do now. Akron, it its current density, might just support one then.
There are almost no benefits of city life that aren’t solely based on travel time and thus eradicated by much faster transportation. Proximity isn’t and end in and of itself. It’s a means to one, which is low travel time. I derive no benefit from having my friends being within 3 miles of me. I derive benefit from being able to go to their house, or meet them at a restaurant, quickly and easily. A friend or restaurant 9 miles away in a Stage 3 world is the same as one 3 miles away now in Stage 1.
Either way I’m excited by the thought that I may live to see a Stage 3 world. That is, of course, assuming I don’t get hit by a bus driven by a human beforehand.
