Artificial Intelligence, Industry 4.0 and UBI | A future without work, or resigning to jobs without a future?
We call it Industry 4.0, it would be the fourth industrial revolution, the first one where the focus will not be on the increase in people’s productivity for the amount of time, but rather the increase in general productivity for the same amount of people. Moreover, the progressive reduction in the number of people will be part of the focus. But, on the other hand, this will probably also be the last industrial revolution we will witness.
Oh I laid down your railroads, every mile of track.
With the muscles on my arm and the sweat upon my back.
And now the trains are rolling, they roll to every shore
You tell me that my job is through, there ain’t no work no more.
Though I laid down your highways all across the land.
With the ringing of the steel and the power of my hands.
And now the roads are there like ribbons in the sky,
You tell me that my job is through but still I wonder why.
For the wages were low and the hours were long
And the labour was all I could bear.
Now you’ve got new machines for to take my place
And you tell me it’s not mine to share.
Though I laid down your factories and laid down your fields,
With my feet on the ground and my back to your wheels.
And now the smoke is rising, the steel is all a-glow,
I’m walking down a jobless road and where am I to go.
Tell me, where am I to go.
(Phil Ochs: Automation Song )
“The next 30 years will be suffering: this revolution will bring social instability, like all revolutions,” says Jack Ma, Alibaba’s founder. What are we talking about? We are talking about the widespread fear that the next industrial revolution – now at the door – will bring automation on such a large scale that it will demolish the labor market.
But we have already been there … have we not?
This is not the first industrial revolution (obviously, since we are talking about the fourth one you say).
In general, the concept of industrial revolution indicates a period of demarcation, due to technological innovations, between two historical phases. Historically speaking, we can say that we have gone through two industrial revolutions, which introduced mechanization and economies of scale. While we are experiencing the third – that of computers and automation – right now, the fourth, that of Artificial Intelligence, is upon us.
The first three industrial revolutions: from mechanization to the information age
The first industrial revolution, between the eighteenth and nineteenth centuries, for example, marked the transformation of western society from almost entirely agricultural to industrial, thanks to the invention of steam engines, mechanical spinning machines, and railways. This automation contributed strongly to the creation of the middle class, which is somewhat the foundation of today’s capitalist economy. On the other hand, however, it inevitably led to the elimination of many jobs.
These job cuts were not taken well initially, generating protest movements like Luddism. However, it soon became evident that the labor market was also transforming in the medium term. The drastic reduction of the hardest jobs corresponded to the creation of new jobs that required expertise and specialization. Substantially more and more manual work was left to the machines, and more and more new jobs concerning the same machine’s management were created and available to the people.
The second industrial revolution is even broader, with many innovations during the nineteenth century, such as stainless steel and anesthesia, electrical lighting, telephone, wireless telegraph and petrol engine, not to mention the alternated current. But above all, it was the era of mass production, thanks to the assembly line introduced by Ford. These innovations radically transformed society, leading to the demographic explosion in the cities, the birth of an economic system based on capitalization and consumerism, and a series of social struggles that would lay the foundation for two world wars.
On the other hand, however, as it had happened for the first revolution, a period of instability followed (at least in the most industrialized countries) a period of great economic growth. Again, many of the hardest jobs were disappearing, with the emergence of new and more specialized professions that substantially contributed to the further expansion of the middle class. But if the first two revolutions had laid the foundations, we can say that technological growth, although accelerating, was still relatively linear.
The invention of the transistor in 1925 and then integrated circuits gave way to a race for miniaturization that suggested to Gordon Moore the formulation of his Moore Law (more of an empirical observation rather than a formal law). According to Moore’s law, in the production of processors, the number of integrated transistors doubles about every 18 months. The result is that since the invention of the phone we have moved on to the Moon in less than 100 years, and then from the first microprocessor to smartphones thousands of times more powerful than all the technology used for the first space missions. Just twenty or so have passed from the Internet to the first autonomous car …
The information age and the Internet is practically the third industrial revolution we are experiencing now. An era that has marked an evolution of the labor market, with an exponential growth of technical professions in the last 10 years.
The fourth revolution, the illusion of “we have been there already” and the paradox of the draft horse
The most common question at this point is “We have been there already, industrial revolutions have always eliminated jobs, but they have always created new and different ones, so why bother? This time will be no different”.
The short answer is that this time it’s different, and the next industrial revolution will probably be the last one.
The longest answer is that if we think about it, seen from the point of view of the draft horse, industrial revolutions have determined the total elimination of all “jobs” and not a single new “use”. Draft horses are of no use anymore, so why should we?
Why is this different? Basically, the differences compared to the previous revolutions are two:
- Machine learning: in the past, innovations have produced increasingly sophisticated tools to support us, but this time is not just a matter of complexity, this time the machines can learn to perform tasks by themselves.
- Evolution speed: this time, especially thanks to machine learning, we are about to enter an era of automation in which new technologies and industries are being created at ever-increasing speeds. The speed at which new types of jobs are created, however, is not comparable to that with which “old” jobs are replaced.
Productivity is increasing more and more, costs are dropping ever more, new types of business are constantly emerging but… jobs are not. However, world population keeps growing. It is worth remembering that the whole capitalist social system is based on consumerism: the whole production chain is based on the assumption that there will be someone who will buy goods and services. Without a job today there is no income, and without income, there is no consumption, and without consumers, the whole system risks imploding on itself.
Do you think your work is exempt from being replaced by Artificial Intelligence? Maybe it’s time to reconsider.
You might think that the scenario depicted above is an exaggeration and that ultimately there will always be the need for someone to maneuver the machines, and to teach them what to do, by maneuvering or programming them.
But as I said, this is the age of machine learning, so I introduce you to Baxter, the first general-purpose robot that can learn to perform tasks simply by observing it while doing it.
The idea of manual work replaced by machines is obviously the first that comes to mind, but the jobs we’re talking about here are not exactly the assembly line jobs we’d expect. In this category, we put not only manual jobs like workers and farmers but also service jobs, like clerks, cashiers and truck drivers.
Agriculture: in agriculture, we already speak of “precision agriculture”, where very few people are able to run entire farms. In this sector, the whole chain is automated, from drones that fly over the fields collecting data on the ground, to artificial intelligence that on the basis of that data irrigation maneuvers, tractors, and collectors.
Retail: Amazon has been experimenting with Amazon Go for some time, and Walmart is doing the same with Kepler. We call it smart retail, that is, on the one hand improving the customer experience, but above all optimizing and reducing costs by automating.
Transportation: while fully automated subways and trains are anything but new today, companies like Tesla and Otto are already producing their long-distance driverless truck fleets. In fact, the news of Otto’s first driverless truck (acquired by Uber) to make a delivery in autonomy dates back to 2016 (see below). For its part, Elon Musk announced the first autonomous delivery by one of its “Seeds” as early as March 7th.
Logistics: automating warehouses is now a reality in many companies such as Amazon and Alibaba, robots are able to move entire shelves in an extremely precise, quickly and independently. Even here it is no longer just the usual mechanical arms, but automata able to move in the environment in a totally autonomous way, finding optimized routes while avoiding collisions. Furthermore, last mile deliveries of goods through drones or autonomous robots are already undergoing testing.
Clerical jobs: not just blue collars
The “white collar” jobs are already those that one would expect less, with the traditional concept of “machine”. After all, they require analytical skills, learning skills, precision, ability to interface with suppliers and customers (perhaps automated in turn). “Wait a minute, is not that what the machines do best?” Precisely.
Press: despite recent interesting attempts, we can say that creative writing is still a human prerogative. However, we see that Automated Insights has long since launched its Wordsmith automated financial reporting service, used by Associated Press and Yahoo! In the field of news production, Google has recently funded the Press Association’s RADAR project .
Tourism 4.0: the travel industry is certainly not exempt from the “smart” wave. Specifically, “smart travel” (or “travel 4.0” if you prefer) is witnessing the widespread dissemination of online services and the use of IoT. People are now able to organize travel autonomously, and the agency desk with the operator ready to offer generic packages is becoming obsolete. The same airlines are planning the heavy implementation of IoT signals to improve the experience of airport travelers, while at the same time reducing the need for personnel.
Lawyers : not even attorneys are safe from the revolution. The skeptic might object that we are talking about a job that requires years of human experience and professionalism, not of “robotic work”, how can we replace it with a machine? Well, the trick stands exactly in that “it requires experience” part.
Noory Bechor is the CEO of the Israeli LawGeex , for example, which produces a platform that can analyze contracts faster than any human counterpart. While the mission is to “help legal teams” in their work, one can’t help but notice that the reduction of the necessary human labor force has begun to shrink here as well.
How he went from a corporate lawyer to CEO of an AI company can be condensed into his reflection:
“I worked a lot of contracts for small companies, as well as for investors and multinational companies. For me it was shocking that I had to reinvent the wheel every time I needed to write or review a contract. All those hours of work on this kind of activity were a pain.”
A considerable part of white-collar work requires the memorization of rules, slow learning through exposure over time to different cases and examples. Learning that leads us to develop the generalization and judgment capacity necessary to carry out the work efficiently. The problem is that this kind of learning is precisely where the machines 4.0 are better.
Furthermore, talking about LawGeex, Bechor’s description can be illuminating in his own way:
“You can take a new contract, one you’ve never seen before, read it and compare it with a database of all the similar contracts you’ve seen in the past.”
And, I would say, it can do it much more efficiently than any human. IBM has also been active In legal research for some time, with its chatbot Ross, able to browse through thousands of documents and offer professional legal advice in the field of bankruptcy, intellectual property and employment.
Medicine: Medical diagnosis is another area in which Artificial Intelligence is strong, and one that we were used to considering as a typical human activity.
But, as for the legal sector, a doctor’s ability to diagnose is acquired after years of analysis of medical reports and clinical tests, comparing them with typical pictures of diseases encountered in the past. Well, this too is a task where artificial intelligence is strongest.
That’s not all: a recent article published by Google and Verily explored how to make early diagnoses of breast cancer through AI. A team of researchers from Philadelphia was able to detect with 99% accuracy the presence or absence of tuberculosis analyzing X-ray scans of the chest, using a workflow composed of GoogLeNet and AlexNet. Last but not least is anesthesia, where for example Sedasys produced by Johnson & Johnson, which was able to perform some types of anesthesia, thus saving on the cost of the specialist.
Of course, in the latter case, it must be reported that Sedasys did not have the desired success, leading to abandonment (called for by thousands of specialists “outraged” by the idea) by Johnson & Johnson, officially for poor sales. However, after in-depth analysis, the cause of this “failure” seems to be more the immature state of research (it was back in 2016), than the absolute impossibility of automating the process.
Software development : but if the machines will replace us in so many jobs, there shall always be the need for someone programming them, shall it not? Undoubtedly today we are witnessing an explosion in the demand for data scientists and AI developers, a demand that today still exceeds the offer by far. But it is unreasonable to expect this trend lasting very long, since artificial intelligence is becoming a “commodity” quickly, and soon the ability to work in this area could stop being a discriminant.
On the other hand, the number of applications and Artificial Intelligence systems is growing exponentially, much faster than the capacity we have to train new specialists. This difficulty has brought Google to explore the problem, and (not too surprisingly) has revealed that much of the professionalism of a machine learning developer involves the acquisition of mathematical skills and exposure to a number of models and case studies, in order to develop the required generalization skills.
Furthermore, the process of inception and development of a neural network consists largely of trials and errors on experiments with many models, until credible results are met. Does it look like a pattern we have already seen? Well, it is: even these are skills where machine learning excels, and Google believes that it has drastically reduced the problem of skill shortage with their Auto ML, a platform able to… develop autonomous neural networks.
It is difficult to say where the process can lead, but at this point, it should be clear that any work that can be optimized, which requires calculations, analysis and even decisions can be (and probably will) sooner or later be taken by a machine.
The manager’s job is no exception, especially in the big data era. What does a manager’s job consist of? The answer obviously varies in details and nuances from context to context, but fundamentally most of a manager’s tasks fall into areas such as general office work, budget management, planning, business decisions, problem-solving at various levels.
The first two functions are all too obviously automatable, the last three categories can be traced back to decision making in general. Now, it is curious to note how little we know at the bottom of the decision-making process of managers, or in other words, how managers make decisions. Eccles and Wood, in a well-known article of the ’70s, published in the Journal of Management Studies, thus began in the introduction:
“Decisions are the visible product of the managerial process, yet we know almost nothing about the real time context of managerial decision-making.”
Universal Basic Income (UBI): the solution?
The picture painted so far may seem catastrophic: jobs that disappear in increasingly massive quantities, replaced by machines, fewer and fewer new jobs created, consumerism doomed to disappear due to explosive unemployment, a society that ultimately collapses on itself.
However, this dark scenario is not inevitable if the society takes the threat seriously and manages to reorganize itself in an appropriate manner. Less work does not necessarily mean more poverty: the solution could be to create wealth for machines, in addition to simple goods. The current capitalist society is based on consumerism: without work today like no money, no consumption, and without consumption everything stops. But it is the first assumption that could be changed, that is with the introduction of the one known as Universal Basic Income (UBI).
The UBI is a concept that recalls the income of citizenship that is often heard of, whose fundamental point is to be unconditional. Unconditional means that it is not linked to unemployment or social or economic status: it is given and nothing else.
But this way there is the risk of encouraging people to turn idler?
Actually no, several experiments taking place in Europe have shown that with a basic income behind them people are more encouraged to invest in training, to try new ways such as starting their own business, thanks to having covered shoulders in the event of bankruptcy. In fact, the sour point is precisely the risk of failure, which in today’s society can only be tackled by very few.
In practice, so far the results say that basic income is seen from people more as an opportunity to find their own way, rather than as free money to play online. Opportunities that would not exist in our current society, where people work more than 40 hours per week do not have time to try anything else and obviously can not afford to leave their jobs to do so.
What about unemployment subsidies?
Well, unemployment subsidies aren’t really a solution, not only because not every country has those subsidies in place, but mainly because that system forces people to take whatever job is offered, or they could lose the subsidy. In addition, the subsidy is terminated (obviously) as soon as the recipient manages to find a job by himself, which is not something that stimulates people to keep looking.
But providing basic income is expensive, how do we find the funds?
Many ways have been proposed, from the rationalization of public spending to the fight against tax evasion. But above all, the most sensible process could be to have this income financed by those who own the machines, and therefore the production. There are not many alternatives: if nobody is able to buy the goods, the cycle is interrupted, and the production becomes an end in itself. Ford himself in the fifties realized that consumers should to be able to afford to buy the cars he produced.
Industry 4.0 is here, jobs will not go away overnight, but the process has started, and is clearly irreversible. I can not say with certainty whether the UBI would be the solution to everything, but it is clear that the current social structure is not ready to absorb the blow, and a reorganization made with reasonable foresight is necessary.
I like to fantasize a world where no one is forced to do horrible jobs just because “I gotta do something to live”, and wherein every place are found only people motivated to build something. It’s a very… poetic dream, but in the end shouldn’t this be the kind of progress brought by technology?