Automation Revolution: Self-Driving Cars and Their Consequences
Ford, probably the most progressive of the American Big Three automakers, just announced an ambitious plan to have fleets of fully autonomous vehicles on the road within five years. Fully autonomous has a specific meaning for vehicles: Level 4 as defined by the National Highway Traffic Safety Administration. (Here’s an explanation of the levels.) For comparison, the Tesla autopilot features are at Level 2.
There is a whole cascade of unintended consequences that probably will happen as a result of the mainstream introduction of autonomous cars into our society. For starters, we probably need to examine various forms of commercial transportation, such as trucking, public transport, cabs and delivery vehicles. For the companies that operate these fleets, drivers are employed only to serve a specific purpose: operating their vehicles. If they can accomplish this purpose without paying driver salaries, they will.
Impact of Automation
The Bureau of Labor Statistics records about 970,800 truck drivers and 203,340 bus drivers for 2015. They earned an average of $23.33 per hour, which means that they represented a total salary cost (not including any benefits) of over $55 Billion per year. How motivated would transportation industries be to adopt autonomous vehicles in order to shed $55 Billion of cost? Answer: pretty darn motivated.
Of course, if we just count truck and bus drivers, that leaves 1.2 Million people out of work, their jobs replaced with automation. This is the sort of situation that leads to economic, social and political upheaval.
To take it further, understand that this isn’t just about truck and bus drivers. Any job that is sufficiently rules-driven, especially ones that don’t require physical manipulation of objects, can be automated. The technology to do so is right on the horizon — we’re talking about timeframes such as the next 5 years. Loan officers? Gone. Low-end retail? Gone. Here’s a handy calculator to see if your job is automatable. Of course, this is looking at a 20 year outlook, but you can probably extrapolate that if a position has a very high (90%+) likelihood of being automated, we won’t have to wait 20 years for it to happen.
A Revolution Brewing
This is big. Potentially it exists on the order of: this breaks the economy. Currently, the way our economy is set up simply doesn’t function when you automate away most of the jobs. Massive economic shift has happened before in history, of course, and when it did, it was accompanied by the word revolution.
The most recent one of these was the Industrial Revolution, which kicked off in the 19th century and upon which our society is still modeled. Before that, we lived in an agricultural society.
The Industrial Revolution introduced specialization. Industries creating specialized goods at scale, employing labor to assist in the creation of those goods. Even our modern service-focussed industry is derivative of an industrial economy. If you follow the money flowing to any web designer or barista far enough, it leads to the industrial manufacturing of goods. Somewhere a thing was built, or something was mined out of the ground, or some energy was generated, and the cash started to flow. The Industrial Revolution brought us employment, and still drives employment in our society today.
But when many (most?) of the jobs we know can be done faster, better and cheaper through automation, what happens?
Possible Futures
The optimistic perspective is that other jobs will appear to take their place. This has happened many times during the Industrial period: while there’s little call for steam engine or typewriter experts these days, if you’re a data scientist you should be able to get a job any time you want one. And to be fair, plenty of so-called “revolutions” that have occurred during the Industrial period have simply been disruptive shifts that resulted in new jobs replacing old ones. If this scenario describes what happens, we should count ourselves relatively lucky.
(Even if it is just a shift, however, there can still be economic upheaval, as it is extremely difficult to retrain large sectors of the working population. Inevitably people will get left behind.)
Less optimistic scenarios postulate that inevitably there will be less employment than there was before. Perhaps a lot less. This is scary, because mass unemployment will tank the economy, and inevitably lead to political and social unrest. If we automate everyone out of a job, we’ll kill the Industrial Revolution’s golden goose.
Some proposed solutions to this scenario are basically, well, socialism. The idea of a universal basic income is essentially a central economic planning-style response to mass automation and the associated unemployment. In essence universal basic income is the idea that you will receive a check every month to cover your necessities. AI may have taken your job, making you redundant, but here’s a check anyway so you can live.
While the idea of universal basic income seems sound, I have trouble believing it to be a politically viable solution, at least without an entire generation of politicians dying off first. Additionally, I think that the major impact of automation will start to be felt long before a government led central response could be mustered.
The Genie is Out of the Bottle
If it’s such a problem, why not just hold back automation then? Stop developing AI? If there are so many concerns about this technology, ranging from AI taking our jobs all the way to a super intelligence-led apocalypse, perhaps we shouldn’t be pursuing it.
The answer here is that the genie is out of the bottle, and it’s not going back in. The only cases where we, as a society, have successfully restricted technology (for example, nuclear fission) have been when we’ve been able to correspondingly restrict access to necessary physical materials (for example, plutonium).
AI has no specialized materials. The algorithms are out there. The means of data collection (thanks Internet) are out there. The motivations for building these systems are clear (better, faster, cheaper). Automation is coming, whether we like it or not.
And, let’s not forget, automation can provide many benefits. More automated cars could lead to less traffic accidents. Automated management of the energy grid could lead to less consumption of power. Automation could provide people with goods and services that they are simply unable to get right now. It can be easy to dwell on the problems of automation, but ignore the significant benefits.
The End of the World as we Know It?
So, are we, as a society, screwed? Well, maybe not. We should first remember the inherent limits of predicting the future: first, it’s really hard to predict how fast changes will take place, and second, we’re looking at the future through the lens of only the ideas we have available today, so it’s entirely possible there’s an aspect of all of this that we can’t yet see.
The timing question is important, because it’s what controls how much time society has to react to change. Change that happens suddenly, will tend to create crisis. Change that happens over the course of decades provides time for society to adapt to it. Ford announcing a five year timeline for autonomous vehicles does not necessarily mean that driving for a living will disappear within five years. Perhaps it will take decades, giving time for the workforce to transition.
The idea lens concept is also critical. Thirty years ago we would never have imagined there would be hot jobs like user experience designer, data scientist and social media coordinator. What new jobs will come to exist that we can scarcely imagine now?
What about the idea of a behavioral designer for these autonomous systems? Someone to determine their personality, how they act in various circumstances etc. The more of these systems that are introduced into the world, the more we’ll have need for someone to design how they act.
How about an autonomous system doctor? Someone who can diagnose and correct malformed behavior in an AI, perhaps caused by training on bad data. We saw earlier this year how Microsoft’s Tay bot was turned into a hate-spewing Nazi by trolls on Twitter — what if someone could come along and cure her?
Taking this more optimistic position requires a little faith: we need to trust that new opportunities will emerge for individuals, and that we’ll be able to adapt as a society to the changes that automation will bring.
(Update: Uber’s Self Driving Fleet Arrives in Pittsburgh This Month. Here we go…)
