Industrial Revolution 2.0
There has never been a time of greater promise, or of greater peril. -Professor Klaus Schwab, Founder and Executive Chairman of the World Economic Forum.
After our blog last week exploring the Russian manipulation of social media through algorithms, I want to explore Artificial Intelligence (AI) a little deeper. The societal impact of AI is a game-changer not dissimilar from the Industrial Revolution, and if we don’t know how to play the game, the effects could be disastrous.
A few weeks back I grudgingly read an article the group Anonymous penned on AI. I’m not a big fan of the group as I’ve seen first hand how their hacking has exposed human assets, endangering their lives and those of their families. Like Anonymous, I’m a big fan of transparency and uncovering “truths” but do believe there are no-go zones to protect national security.
I also haven’t paid much attention to Anonymous’ focus on the 99% v 1%. Partly, it is out of laziness. Partly, I just really don’t identify with this movement. I recognize that there is a group of American “oligarchs”, a small group of ultra-wealthy that have amassed considerable governmental sway, and our Gini coefficient — the measure of the wealth gap in countries — has grown to reveal this staggering divide.
I am definitely in the 99% but gratefully, all of my basic needs are met and then some, and while I do like money, it’s just not a priority. Security and comfort are, but wealth for wealth’s sake, not so much. However, this article struck a chord. I’m kicking myself for not saving it and I can’t find it again, but the basic takeaway is: the tech gurus building AI will continue to get richer while our country, as a whole, gets poorer as AI replaces workers. This is the stuff of revolutions.
I just can’t keep my head stuck in the sand anymore.
Paul recently shared his thoughts on AI and did such a lovely job, that I am going to publish it wholesale. He gives just one example of AI — self-driving cars — and you can see the impact. Imagine this combined with AI in other industries.
Here’s Paul.
Driving is one of the largest employment sectors in the U.S. It is especially important for male employment. Truck drivers, taxi drivers (Uber drivers), chauffeurs, delivery drivers, postal drivers, etc make up a big portion of the workforce. Also of course, there are all the truck-stop workers, hotel staff, food and beverage workers, insurance workers etc who serve them. Globally, driving is a huge portion of employment, and may be as much as 20% overall for males. Also, beyond that of course, a lot of people work in traffic control, road safety, traffic policing etc. Here is a map of the most common jobs in the US by state:
Also significant is that truck driving is actually quite a well-paid job, given the technical skills needed to perform it. It is thus a financial lifeline for many families.
The problem is, us humans are not very good at driving — not to say that truck drivers are not, they are probably, given the amount of time spent on the road, among the best — but the changes that are coming will affect everyone. As a whole, humans don’t drive so well. We get tired, we get distracted, our reaction times vary with food, age, sleep levels, stress etc. Our brains miss information or interpret information incorrectly. We fail to see cyclists. We misjudge the speed of other vehicles. We reach down to adjust the radio and miss the turning truck. We answer the phone, we text, we turn around to stop the kids arguing in the back seat. We drive after drinking or taking drugs. The list goes on.
In 2016, 37,461 people died in traffic / highway incidents in the United States. This included more than 6,000 pedestrians. The number of people injured was several times this total. Altogether, this toll of death, heartbreak, trauma, and huge medical costs is extremely high. This is obviously many times the number of people killed in terrorist incidents in the U.S. in decades.
In addition, the fact that we need to drive our vehicles with humans means that vehicles need to be parked near those humans. There is no point having a car if you need to travel 10 miles to get to it. So, cars are taking up valuable space all along our streets, in our cities and on our properties.
Self-driving (AI-drive) cars do not get tired. They do not need to stop for the toilet or food or sleep. They don’t drink and drive. They don’t suffer from road rage. If they are wired up to good sensors (including networked sensors from other vehicles and even roadside / satellite streams) they can get a much better picture than a human of the surrounding environment of vehicles, pedestrians, cyclists, and obstructions. Their reaction times are potentially many times quicker than ours, and they can process all the data as opposed to our brains’ ability to just filter things that seem important. They could share information with millions of other cars and sensors — thus constantly upgrade their driving abilities in real time — as central repositories automatically analyze the data coming in and communicate the changes back to the fleet.
At the moment, Google, Tesla, Daimler, Uber, Mercedes, Toyota, BMW, and several Chinese brands are working on AI cars. The safety records are very impressive (the Google AI car crashes are all human caused so far, although there is some controversy around Uber). What if AI-driven vehicles offer us the chance to cut the death rates by 50% or 90% or 100%? Tens of thousands of American lives could be saved every year. And it’s not just safety…
Compared to human driven cars, AI driven cars / trucks / taxis / buses:
· Don’t need to be paid. Those high salaries to drivers can be saved by the employers, cutting costs for consumers.
· Can be parked somewhere else. Your car drives you to your office, then drives itself to the outskirts of town to wait in a huge car park, then returns to get you to take you home, then goes back to the huge lot to wait, then comes to get you to take you to the baseball game, then goes to wait, then comes to get you, then takes you home, then goes to spend the night back out at the big lot. The extra energy may not be an issue if this is an electric car sourced from renewable generators.
· Will not speed, be used in illegal crimes (unless they are hacked), flee from police etc. Police may even be able to force a car to pull over and park, as well as being able to see who is inside, whether they have weapons, but NOT whether they are drunk, because that probably won’t matter much.
· Can be used to pick up the kids without the parents spending the time to go and sit in traffic, meet them, and bring them home.
· Allow the people inside to do other things, e.g. watch movies, do some work, read, chat with friends on video, etc. Commuting to work suddenly becomes an actual part of work, or an actual part of rest / relaxation, not something in between.
· Can react to the data environment for traffic, roadworks, congestion flows, and times. This can optimize travel times.
· Allow children, too young to drive, but especially the elderly or physically disabled, who may be unable to drive safely, a lot more independence. Visit friends, visit family, go to events, without having to rely on parents / children to drive them.
So, if we imagine that we reach a future where AI vehicles are as good as I describe above, which seems at least possible, and that we can reduce traffic fatalities by say 90%, or even just 50%, can we morally justify not adopting them, and perhaps even legally require them to be adopted?
Can you imagine the headlines where nearly every human-driven car fatality can be accurately described as unnecessary?
But, what about the employment situation? What about all those families whose comfort and ability to live good lives depend on one member being a truck driver, or taxi driver? Millions of American families would lose their primary source of income. How do they survive?
And, AIs such as these may soon be deployed in a whole range of fields, not just driving. They already operate factory robots, crunch medical data, crunch population data, work out which adverts to show to whom, work out which offers to show which people when they come to the supermarket, route electronic communications, and a lot more.
Pilots may also lose their jobs, as military pilots are already being replaced by drone operators. Autonomous oil tankers, cargo ships, and ferries may be coming too. There is a potential huge economic shock coming, and it may come very quickly. As usual, it looks like it will favor the “already wealthy”, since the trucking companies and their shareholders will save money, while the truck drivers will lose their incomes and medical insurance, etc. Those with the capital to own the AI cars and stock, but more importantly the operating system IP, will reap the benefits.
It is likely that households will still buy AI cars, but it is actually no longer necessary. Hiring a car for a few minutes, a few times a day, with no driver to pay, may make more financial sense than owning an AI car, so the car companies may just set up AI car rental networks in each major population center.
Is this just another worry, like so many in the past, about technology?
People were very worried at the beginning of the industrial revolution about the job losses. Are people worried about rising AI just “luddites”? Maybe, but maybe not.
Look at the depressed industrial areas around Detroit and Flint. Were they wrong to worry that the car industry jobs were going (granted in this case partly to China / Japan as well as to automation, but the effect may be the same as if AIs “did it”)? Did we react well to ensure they could continue with their futures and take measures to protect those communities? Are we adequately prepared for all the AI related disruption that is almost here?
If you think that we should restrict the technology to protect the livelihoods of drivers, then you need to think hard. Should we stop toothpaste / tooth cleaning technology so that the dental industry can benefit from the extra work that they had lost? Should we restrict preventative heart medicine so that the heart surgery industry can hire more people? Should we ban Microsoft Word so that typing secretaries can regain their role in the workforce?
How do we protect against the disruptive economic shocks of AI job replacement that may be coming our way?
· Universal basic income is one often discussed method and is even favored by some libertarians. But who pays for this? Do we tax those generating AI? Is it fair to tax average middle-income earners the same as the AI industry leaders benefiting from the AI revolution?
· Citizen collectives of capital ownership — for example, could current truck drivers insist on being given shares in trucking companies, before they lose their bargaining power? Could they “buy individual trucks” that are then networked into the operators’ companies, the same way taxi drivers invest in the medallions in New York, or the same way that passenger airplane leasing works? They need to act fast, and I see no sign of them taking such steps in time, after all, they may only have a few years — Tesla and Daimler are already road-testing self-driving trucks.
· Radically expanded social welfare — this may be possible in Europe but would require a political and cultural sea-change in America. And, there is a downside to welfare as already noted in its current more limited state. What is the impact when welfare is doubled or even tripled?
· Phased Introduction — for example, limiting technology uptake to give society the time to adjust. First step, no new truck drivers from now on. Then whenever truck drivers retire they are replaced by AI. younger drivers can maybe be brought into a collective ownership. Still, how do you justify the unnecessary deaths caused in the meantime? The families who lose loved ones in human-caused traffic accidents?
So, what do you think? Does the benefit to human life (literally), outweigh the massive social dislocations? The industrial revolution, in part, kick-started the wealth gap. AI has the potential to obliterate it almost entirely leaving only the haves and the have-nots if we don’t pre-empt it with sensible policy solutions.
As I mentioned in our blog last week, I’m no fan of regulations. I’m a big fan of Adam Smith and his “invisible hand” theory, but I think it’s fair to say, Adam Smith most likely never envisioned this. Given these predispositions, I like Paul’s “citizen collectives” suggestion, even though it would entail government regulations to mandate. However, it seems like the best solution with the least regulation.
I told a very progressive friend last week after doing this research, AI may be the only thing that has so far shifted my conservative bent. Behind all the noise generated between the progressives and conservatives over our current administration, there has been very little discussion on how to manage AI. The debate over bringing manufacturing jobs back to the U.S. may soon be passé.
AI is real. This is happening, and we aren’t prepared.
-Paul Harding & Jennifer
PS: Finn graduated from middle school last week. I gave him one week off…
