The case of the People vs. the machines

In a previous article, I analyzed the progress of various technologies, concluding that Artificial General Intelligence (AGI) was still a long way off. There are, however, many areas where technology is soon going to have a tremendous impact. Many jobs, and perhaps entire industries, are threatened by machines and software.

A study from Oxford estimated in 2013 that nearly 50% of current jobs would be on the line by 2033, confirming that this threat was indeed very serious and very real. This article from the Huffington Post made a good job at summarizing jobs that are unsuspectedly subject to automation.

Fifty percent of all jobs! Yet our politicians and media seem mostly concerned by our merely-stagnating unemployment rate, which is obviously negligible compared to what is, slowly but surely, coming to us. Perhaps the number is too big to be taken seriously. Perhaps the threat is too abstract, and too far away to scare the average person talking on television. In ten or twenty years, they will remember how short-sighted they were today.

It is not too difficult to picture a world that would be 2° or 3° warmer, with 2m or 3m higher sea levels. However, a world with a 50% unemployment rate seems plainly impossible. Not because we do not believe in technology taking our jobs, but because we can all see that society would collapse into chaos before reaching that point. Drastic changes will have to happen before that.

Will technology generate a plethora of new jobs to replace the old ones? Will the governments regulate all the problems away? Will we voluntarily turn our backs on Amazon and buy our books at the local bookstore? Will we get used to learning new jobs every five years or so? Is technological progress a good thing for Humanity? This is the debate of the century.

In the last six months, I found myself debating these questions with many different people, with varying experiences and backgrounds, resulting in widely different reactions and arguments. What is amazing is to see how poorly informed the public is. The French media often depicts the activities of the Silicon Valley with contempt at best, and usually resentment when it comes to them not paying their taxes. They completely fail to see that the world is being drastically reinvented by a few people with a vision ten or twenty years ahead of us.

When the information gap is filled, I find that people generally fit into one of two groups: The enthusiastic and the terrified. Clearly, the eldest most often belong in the second category while those who “get” technology generally belong in the first one. But behind the question of age or skill lies a more dividing factor: the ability to learn. Not everyone has the ability to learn. Those who use the Internet since childhood, who use Google Search a hundred times a day, who have read countless blog posts and tutorials and who commonly follow Youtubers naturally have this ability (and do not really see what the big deal is about it). But many have missed this opportunity: they still do not know how to effectively find information online; they are terrified at the idea of learning a new skill, tool, programming language, etc. They would do anything not to have to learn anything new, in particular they would fight for their job to the death. In the approaching world of perpetual job obsolescence, there is no other way to go by than to keep learning, over and over again. Those who stop learning most certainly have a crisis ahead of them.

I think it is possible to switch from the second to the first group. But the change should not happen under pressure. Rather, it is through education and information that people should be compelled to learn how to learn. Consequently, I propose to look in deeper details at some concrete examples of technologies that will have a major impact in the next decade.

Autonomous Vehicles

Self-driving cars are now well on their way. All automobile manufacturers are fully on track and have all developed functional prototypes reaching different degrees of sophistication. Google was a precursor with their fully-autonomous fleet of cute little cars now legally driving on open roads in California. Tesla recently upgraded their onboard software with a self-driving mode (which reportedly gets smarter the more miles are self-driven with it). Volvo is testing its own self-driving car on the roads of Sweden with plans to sell them to first customers next year. BMW partnered with Baidu (the “Chinese Google”) to develop an autonomous car for the Chinese market. Apple themselves have allegedly started cooking an “i-Car” (So-called project Titan). The list goes on. Even if one manufacturer was too far behind technologically, it could probably just buy the software from another one.

Google’s driverless car

Google’s driverless car gets a ticket for driving too carefully!

This reality is hard to believe for many people. “I would be too scared of riding in a self-driving car.” “What happens legally in case of accident?” These objections are valid, of course, but miss the point that self-driving vehicles will actually end up being significantly safer than today, with reaction delays dramatically reduced compared to human drivers. It is not that machines have suddenly become that much smarter (they do not need to), but rather that human capacities are often a bit overestimated. A crushing majority of road crashes are caused by drivers breaking a rule and could be easily avoided.

At first, self-driving features will only work on highways, where it is the easiest, and will require a driver to take over in urban areas. But, before long, they will manage to get passengers from any point A to any point B. As long as a driver is legally required, the benefits of self-driving cars will be limited to increased comfort and safety. But of course, at some point this requirement will become mostly artificial and technically irrelevant, and the critical step of removing the driver will be taken, unleashing much bigger consequences.

All the road transportation industry (taxis, trucks, deliverers…) will want to switch to fully-autonomous driving, putting millions of people out of a job practically overnight (3.5 million professional drivers in the US). The entire business of car insurance will be rendered obsolete in no time. The potential impact on employment is so massive that governments will probably have to delay this transition, provide welfare and training or somehow force those industries to keep drivers on payroll for some time. Anyhow, these measures should be debated as early as possible between all parties.

On the other hand, many positive consequences arise: Suddenly, owning a personal car will become pointless. Without drivers, the cost of “Ubering” will drop dramatically, and we will make all of our daily trips in on-demand vehicles summoned at our homes and workplaces in minutes. Carpooling will work much better than today as all the constraints of planning and organization will vanish. (Ironically, all this gain of efficiency seems disastrous for car manufacturers)

On the long run, we can even imagine that traffic jams, traffic signs and parking lots (and more generally the concept of “needing a place to park your car”) will disappear. The cost of moving goods will drop as well, acting positively on prices and on the cost of living. The global energy consumption of cars will decrease, as rides will be shared, traffic will be fluid, and automatic driving will be smoother and more efficient than human driving. As car accidents will disappear, so will their associated cost for society (estimated to almost $1 trillion in the US in 2013).

Does this future seem crazy and far away to you? Remember that Google is already operating a functional fleet of driverless car on the streets of Mountain View. It is not a matter of technology any more. Only a matter of time.

Online Retail

Online commerce has gone a long way since the beginnings of the Internet. But if you think that it has reached its pinnacle, think again. E-commerce represents only about 7% of total commerce (in the US). But this figure is growing steadily at an increasing rate of 0.6% every year. It does not sound like much, but every percent represents about 50$ billion (for the US only).

Growth of e-commerce

Share of e-commerce vs. all commerce (note the peaks at Christmas time)

Note that E-commerce is largely dominated by one actor: Amazon, which stays way ahead of its nearest competitors.

Amazon share in e-commerce

Amazon, crushing its competitors

Why this hegemonic position? Because the digital economy naturally tends to favor monopolistic situations. In the physical world, you can choose between various supermarket chains, various phone contractors, various car manufacturers and various clothing brands. But the Internet is a winner-takes-all world. That is why you probably use Facebook rather than Google+, Uber rather than Lyft, YouTube rather than DailyMotion, etc. The main reason for this phenomenon is the law of Increasing Returns. In the physical world, costs (building stores, hiring clerks) stop you from growing. But on the Internet, serving more and more customers costs you little, and the more customers you have, the more data you collect, the more you learn and the more attractive your products become. There is no point in splitting the market with competitors if nothing stops you from conquering the whole of it.

And the whole market is what Amazon is after. If you think that they are big today, remember that E-commerce still represents only 7% of all commerce. The problem is that, every percent they conquer makes it more difficult for physical stores to survive. Every percent lost makes it less logical for brick-and-mortar businesses to invest in real estate, build stores, hire people, manage logistics, etc. Retail stores have a chaotic life and poor economies of scale compared to Amazon who runs gigantic warehouses like clockwork.

But the biggest problem retail stores have is that they cannot fight on the front of customer experience. Shopping on Amazon is just so easy and efficient that when you set foot in a physical store, all the drawbacks become obvious: you have to wait in line to pay, you have to search for products, you never know if the prices are fair, the range of product is too limited, you cannot look at your basket and know how much it is worth, you cannot see what other customers had to say about a product, and so on.

It is possible that in the coming years, the slogan “You can buy anything on Amazon” will gradually turn into “You do buy everything on Amazon”. Of course, there are many areas where retail stores keep a strong advantage: If you shop for food, clothes or furniture, you generally want to feel the product before buying. But is this even such a hard barrier? Nothing is less certain. With food, for instance, the explosion of the “Drive” supermarkets has proven that consumers were ready to trade the pain of grocery shopping for the risk of finding a bad apple in their basket. For clothes and furniture, the development of virtual reality will allow customers to try articles on their own avatar, or visualize a couch in the middle of their living room.

The time when Amazon was only threatening bookstores is behind us. Millions of jobs in the retail industry are on the line, and they are running out of competitive advantages. In a not-so-remote future, opening a physical store will be considered odd, and perhaps later, ludicrous. In fact, for the youngest generations “born with a tablet in their hands”, shopping in the real world will sound like a funny idea, something you do when you cannot avoid it.

Companies with new products to sell will more and more work directly with Amazon Marketplace, to avoid the need of having to manage their own website and logistics. This is good news for everyone: The company can focus on design and innovation, while Amazon takes care of all the distribution, and to some extent, of the marketing. Consumers will have the simplest possible shopping experience, saving a large amount of time and energy.

The only thing that could stop Amazon is a massive boycott of the public: A voluntary decision to renounce the comfort and efficiency of systematic and ubiquitous online shopping to preserve local stores and employment. But note that like other Internet giants, Amazon is a colossus with feet of clay. It works because it makes little-to-no profit. If it was to betray the trust of its customers (say, by increasing prices), it would immediately allow for competitors to step in and divert their customers.

Knowledge Management

For the last twenty years, the Internet has revolutionized our conception and organization of knowledge. Not one day goes by that we don’t google our way to the answers of many questions and the solutions to many problems. People all over the world gather in ever-growing online communities and happily provide their time and wisdom to help complete strangers, in exchange for a bit of self-satisfaction and self-promotion.

Yet, if you ever worked in a big company or a big organization, you probably experienced a very different relationship with knowledge. At first, you probably wondered why there was no internal Google or Wikipedia to answer your constant flow of pressing questions. Then you probably realized that, even if there was, they would be of limited interest, as people would most likely not share valuable information on these platforms.

For a large traditional company, information is gold and people are very protective of it; not only externally but also internally. Interestingly, both because of the same phenomenon: The fear of being out of business. Managers are trained to behave as the CEOs of their own department, with their own budget, headcount and objectives, and as such, defend fiercely the interests of their team, sometimes over the interests of the company. In particular, they regard knowledge as a competitive advantage (to access more budget, more influence), and think twice before answering a question that could reveal their own secret sauce. Obviously, this secrecy generates inefficiency, tensions and unproductivity.

So many industries are stuck in this pattern. You might think that building an aircraft is about technology. It is much more accurate to state that it is about knowledge. A significant barrier to entry that in-place aircraft manufacturer benefit from is the extensive knowledge of how to certify an aircraft with the European and American authorities. Getting an aircraft to fly is relatively easy, but getting it approved for operations is a nightmare, particularly for new-entrants who heavily struggle with this milestone (and sometime go bankrupt).

Note that there are alternatives to this model. It seems that major technology companies, with the notable exception of Apple, are choosing more and more to open-source a large part of their knowledge. The field of artificial intelligence is a good example, as two major leaders, Google and Facebook, both decided to share their software (their primary asset!) and publish their studies with the rest of the world. In 2014 Tesla even open-sourced all their patents, to promote the development of electric vehicles (full announcement from Elon Musk). These companies are no charities: They do it because it is good for business. In particular, they gain influence in their fields, hire top brains and get to impose the dominant standards to the other actors.

But even in traditional companies, where knowledge is secured and carefully traded, the revolution is near. With the rise of Deep Learning and Natural Language Processing (NLP) frameworks that get closer to human-performance every year in many areas, a major productivity step is merely awaiting to be taken. When all the documents and emails of a score-old company go through algorithms such as Google’s SyntaxNet, these companies will finally get their own Google-Wikipedia-Siri application. The effect will be incredible. What happens when most of the bottlenecks in a company are suddenly removed? What happens when engineers from a knowledge-dominated industry are suddenly competing with a digital, Internet-fast version of themselves?

The next question of course is: what does a traditional company do with a massive productivity boost? It can either choose to develop more products or to cut into payroll at a constant production. The latter is more likely to be chosen, simply because it is so hard to train traditional engineers to do something completely new.

The End of Work (as we know it?)

As more and more intellectuals and economists are claiming, it is likely that the current dominant form of work (9-to-6, Monday-to-Friday) is going to become an exception. The most precarious jobs are already going through a large-scale transition with the so-called “Uberisation”. Soon, high-value engineering jobs will go through the same path. Note that Uberisation itself is actually is actually rather soft, as it only transforms jobs, without destroying them.

Far from being tragic, it is good news for dynamic workers, eager to work and learn new things. They will realize that they have most likely been under-payed in their traditional company. This is bad news for rent-seeking workers.

The digital economy just does not tolerate unfair rents and monopolies. This does not only include obvious rents (French taxi drivers, notaries, air-traffic controllers…). It also applies to many workers who have managed to build and protect their own territory inside their company by keeping a monopoly on a particular skill or piece of knowledge. In fact, it rather applies more to this latter category of workers, as real rents will be hard to get rid of (as they can basically prevent the country from functioning by just going on strike – which they claim is proof of their usefulness).

Note that consumers do tolerate and even encourage the monopolies of the GAFAs (Google, Apple, Facebook, Amazon), but only because they make relatively small profits, compared to the tremendous value they bring to the world. If Google employees somehow decided to go on strike and put Search offline, then the users would just grudgingly move to Bing.

As traditional companies turn into digital ones (or are eaten by them), their business model also needs to evolve. The days of the knowledge monopolies protecting a rent and preventing competitors to enter a market are numbered. Like for any good digital company, the only way to survive is to innovate and grow. The traditional worker used to ask “how can I protect my job?” while the digital worker wonders “how can I automate mine and get a new one?”

The optimistic observers hope that the digital revolution will bring a whole lot of new jobs to replace the old ones. I personally do not share this view. Surely, the Internet has created the job of website designer or community manager and we are currently seeing an explosion in the numbers of data and AI scientists. But clearly, the goal of computers has always been to perform work instead of the humans. Building a website was a difficult thing twenty years ago. Nowadays it is matter of a few taps and clicks. Additionally, it is hard to imagine that we could convert forty year-old truck drivers to programming, unless they prove to be extremely interested and motivated.

Influencers from the Silicon Valley clearly share the dream of a world without work, where driverless cars pick you up anywhere and your food is delivered by drone every day, with no human in the complete supply chain. We are far from this world, but the advances in technology are surely going in this direction.

In the last century, the extreme gain in productivity brought by technology has had various effects on the job market. One of them is the creation, by capitalism itself, of the famous “bullshit jobs” (This famous article is a must-read). Bullshit jobs are those supposedly-intellectual jobs that are not really useful to society and do not make workers happy, but are paradoxically expanding over the years. Bullshit jobs are often caused by the snowballing effect of increasing complexity. As legislators pass more laws without removing old ones, more lawyers have to be employed; more engineers are needed to design products that fulfill all the standards; more managers are needed to manage those lawyers and engineers; etc.


The phenomenon of Bullshit jobs, brilliantly depicted in the classic “Office Space”

It is therefore a possibility that all those jobs (truck drivers, retail sellers…) will be replaced by more bullshit. However, the rate at which those jobs are going to be destroyed is unlikely to be matched by anything (especially now that the legislators are realizing that regulating the problems away has the harmful effect of building or consolidating rents).

It should also be noted that digital companies have a particular hatred for bullshit. In digital companies, whatever can be automated will be automated. This means in particular that managers are actually doing management, and not spending their time crunching Excel spreadsheets and filling forms.

Good or Bad?

All in all, the fact that we will work less should be great news, as we will be creating more wealth with less effort. Consumers will buy better and cheaper products. Workers will perform more interesting and intellectual tasks. But how to manage those whose job become obsolete, and who do not belong to the technology-enthusiastic category? In a country plagued with unemployment, it is hard to hear “you will just learn something new”.

The inevitability of job destruction conflicts strongly with the leading paradigm of most Western countries: Keeping unemployment as low as possible, even at the cost of subsidizing wasteful industries, passing absurd laws and protecting monopolies. This century-old idea that everyone should work might just be coming to an end. Why should everyone work in a world where so much wealth can be created automatically? Why should we keep creating increasingly absurd and unfulfilling “bullshit jobs”?

A reasonable solution to this conundrum (expectedly promoted by Silicon Valley influencers) might be the Universal Revenue (or inverse tax). It is what it sounds like: Giving everyone a fix and unconditional revenue. Although this could be the subject of an entirely new article, I find that this is the most realistic and exciting solution at this point. Many people remain skeptical that this could work. But I find that these are often the same people that are skeptical of technological progress. Perhaps they are just waiting for their job to be “Uberized” to realize how relevant a Universal Revenue could be.

Gaspard Koenig

In France, Gaspard Koenig is the most active advocate of the Universal Revenue

In a coming article, I will try to depict what would a society without the pressure of having to work look like. A society where materialism would be slowly replaced by unmonetized and abundant knowledge. Some might call it a “Society of the Mind”.