Trend Analysis "Artificial Intelligence (Part 2)": Will Computers Take Our Jobs?
First, please enter the information requested below. The form must be filled out completely. Then, simply click on the "Download" button to download the file directly from our website. We guarantee that the personal data you submit will be used for internal purposes only, and will not be shared with any third parties.
It doesn't often happen that my trend analyses are overtaken by reality in just eight weeks' time. Precisely this, however, has happened with my last trend analysis "2016 Will Be the Year of (Artificial) Intelligence." There I had written that the intelligence of computers has, in fact, achieved superhuman levels in many disciplines, but remains at the 6th Dan when it comes to the royal game of Go. If computers would continue to improve at projected rates every year, I wrote in my prognosis in January, then they would still need ten years before we would see the first computer world champion.
There you can see that even futurists are not shielded from choosing our very human culture of linear prognoses ... and thus ending up completely wrong, because digital technology develops at an exponential rate. A few days ago, the computer "AlphaGo" cleaned house with the world's best player, South Korean Lee Sedol, in a 4:1 series. The human had no chance.
It is in fact probable that computers will not only become better than humans in specific games and isolated areas, but that they will achieve the general intelligence of human beings. And, it is even more probable that the increasingly faster express train of artificial intelligence will not stop at the station "human intelligence," but will simply keep on going. Then, only a brief time later, we will already see computers with superhuman general intelligence. And then things will really get exciting, because intelligence does not automatically imply reason. At least not reason defined in terms of human values.
For this reason, today I would like to continue our small series of trend analyses on the "Future of Intelligence." While in our last issue I described the state of computer intelligence today, how we got there, and why 2016 will see a new breakthrough, today I will focus on the questions:
What further development will we see in our lifetimes?
And will computers take our jobs?
According to our prognoses, the clear answer to the second question is: NO!! Modern post-apocalyptic scenarios are hot sellers, but lack any sort of real foundation. We continue to be on track towards the world of full employment. But read on for yourself ...
The next ten years
Let's examine the near future first. Are you familiar with the Turing test? The test dates back to the year 1950. Practically ancient. Back then, the British mathematician Alan Mathison Turing suggested introducing a test for determining the point in time where artificial and human intelligence had become equal. In the test, a human subject holds conversations with two partners using a keyboard and computer screen. One conversation partner is human, the other is a computer. From that point onwards where the subject can no longer determine which partner his human and which is a computer, it will have been proven that artificial intelligence had achieved parity with human intelligence.
In the next ten years, computers will pass the Turing Test in most individual areas. This will not mean that they have achieved the level of human general intelligence, but it will meant they have come a step closer. More than anything, this will mean that we humans will come to trust in technology more and more – until we trust it more than other human beings.
Thus we will also get used to the fact that computers will understand – and use – all five human senses. Note that this does not mean only hearing and sight, but also taste, touch, and smell. And digital devices will come to understand even human emotions, thoughts, and of course language – and will express and react to these themselves. This will lead us to interact with them in very human ways.
No small number of the world's futurists believe that some of the world's greatest problems will be solved by intelligent computers by the year 2029. They will help us to manage climate change, to establish a regenerative energy system, to democratize education, to prevent terror attacks, and to achieve scientific breakthroughs in medicine and genetics.
Thanks to intelligent computers, we humans will live longer and healthier lives. The reason for this will be the constant analysis of ALL body data. This will lead to recommendations for medical nutrition, healthier behavior, and personalized medicine. In genetics, the costs for the complete analysis of the individual genome will sink drastically, leading to completely new therapies and possibilities for treatment.
And, not least of all, these intelligent computers will become invisible. They will always surround us, but we will no longer see them, because they will be part of the basic fabric of our lives. They will be integrated in virtually every object around us. Intelligent computers require even more data transparency and data exchange, and will make every object not only more individualized, but also adaptive. Products and services will adapt to individual users, but also to their changing usage situations.
Will we all lose our jobs? On the contrary! Full employment by 2040
It is no secret that, even today, computers and their robotic helpers are better workers on production and assembly lines than their human counterparts. In the years to come, they will also become the better knowledge workers, the better doctors, brokers, travel agents, teachers, tax advisors, interpreters, salespeople ... and so on.
Nevertheless, we futurists continue to predict an era of full employment for Germany in the next 20 years.1 The reason is simple: In Germany, digitalization goes hand-in-hand with a unique demographic trend: The huge masses of workers from the generation of the baby boomers will enter retirement by 2025. This number will only fractionally be replaced by the few workers emerging from the following low-birth generations. Anyone who compares retirement and labor market entry figures alone will quickly realize that we will have some 6.5 million fewer working individuals in the German job market.
If we then deduct today's unemployment figures and all the aid programs from the state and the corporate world, this still leaves 3-4 million unstaffable jobs. Digitalization will not be able to replace this many jobs even in ten years. We futurists estimate a maximum of one million automated jobs by 2025. This leaves a situation of full employment that will last until roughly 2035.
During this latter phase, digitalization will supplement human labor. It will make it better, faster, more perfect. It will cause the standard segment to disappear from all markets, changing the traditional market pyramid with its clearly defined economy, standard, and premium divisions. By 2040, only two viable segments will remain: economy and premium.
By 2040: The disappearance of the standard segment
The logic behind digitalization shows a shift in human values. Previously, seemingly all segments from economy to premium functioned according to the same rational logic: the value-for-money comparison. Correspondingly, we identified low prices and low quality in the economy segment and the highest prices and quality in the premium segment.
Logically, this left a huge standard segment in the middle, with middling prices for middling quality. But this paradigm no longer exists!
The most important future development leading to 2040 will be that the economy and premium segments will come to function according to different principles. While the previous weighing of price against quality in the economy segment will remain and, within this segment, will lead to highest qualities and highest prices, customers in the premium segment will decide for products not according to quality and price, but rather according to how these products are suitable for the purposes of identity management.
Because the economy and premium segments will no longer function according to the same logic in the future, there will also no longer be any basis for an intermediary standard segment. The result: It will disappear. The business models of the future will take into account this foundational transformation of the market pyramid and will have to change their sales strategies accordingly.
By 2040: How economy and premium will work
In the next ten years, technologies for intelligent prognostics will make their way into our sales processes. At that point, digital assistance systems will become intelligent. These will be tailored to the individual user and will be located on the customer's smartphone display. Using data analysis as a foundation, they will understand how their users tick and what customer needs are driving them. Additionally, thanks to the basis of situational data, they will understand how their users' consumer needs change from moment to moment. These systems will be part of a large "intelligent touchpoint management" system that will combine the huge amounts of data gathered by many different digital devices, and from this analysis will draw intelligent conclusions. The recommendations then given by these devices will thus not only differ by user, but also with every usage situation. Taken together, this means "adaptive" solutions.
As a result, sales processes – but also products and services – will become adaptive in order to best fit the current usage situation of the individual. Additionally, customers will experience that their digital assistants will make much better offers than conventional sales staff. Users will even come to trust the recommendations of their digital assistants more than their own personal searches, because the answers the assistants give regarding questions of price and quality – questions that can be calculated rationally – will simply be better hands-down! The economy segment of the year 2020 will be characterized by intelligent mobile devices, passive customers, and adaptive offerings.
The premium segment will be the opposite of the economy segment. Here customers will consciously refrain from seeking the rational answer of the best quality and price. They will simply turn off their assistants at this point. But why should they do that when these systems will be so intelligent?
The answer is: Because in certain consumer situations, human beings do not act according to rational logic, but according to the logic of identity management. This means that they make decisions whose chief aim is not the fulfillment of rational necessities, but rather the expression of personal identity.
In these cases, people want to prove to their fellow human beings, their coworkers, their friends and acquaintances – and naturally to themselves as well – that they are special: especially eco, especially cultured, especially athletic, especially innovative, especially frugal, especially luxury-oriented, especially style-conscious, especially tied to their home region ... there is a huge (albeit finite) number of possible identities. In the premium segment, the driving customer need is to go to locations and to buy products where one's identity can best be expressed.
All of this is changing business models, products, and services. But it will not eliminate jobs for human beings.
After 2040: Economy jobs vanish! The prognoses through 2070
The wheel of history, however, will turn again. In my first trend analysis in this series on artificial intelligence, "2016 Will Be the Year of (Artificial) Intelligence," I described the probability of the emergence of human-level machine intelligence, or the equality of artificial intelligence and average human intelligence at 10% for the year 2022, 50% for 2040, and 90% for 2075.2 The average of these prognoses indicates 2045 as the year when humans will no longer be the most intelligent species on earth.
It’s from that point on that things get exciting. One should always be mistrustful of quantitative predictions about the future, because the future can neither be counted nor measured. But some of these numbers can, in fact, show the basic orientation of future developments. Thus, in his report to the US Congress in February, President Obama cited a study conducted by the White House’s Council of Economic Advisors (CEA)3. This study predicts that all jobs in the US labor market that earn less than $20 per hour will be replaced by computers. At present, this affects 62% of US jobs.
But even scientists by like Moshe Vardi (Rice University, Guggenheim Fellow) estimate that half of the jobs in the US today will disappear by 2030.4 Oxford University cites the figure of 50% between 2025-2035 and McKinsey 45%.5 This means primarily simple jobs: fast food workers, call center agents, supermarket cashiers. Among more demanding jobs earning between $20 and $40 per hour, the probability of replicability drops to 31%, and among jobs bringing in more than $40, sinks even as far as 4%, according to the White House study.
These figures should not necessarily be accepted as absolutes. Several voices find justified grounds for skepticism regarding these post-apocalyptic numbers outside of the US: the German job market, for example, shows a significantly higher average level of qualification. Other trends, such as population decline due to demographic change, were not factored into the studies mentioned. Also, the US has 20-25 years to adjust its labor market accordingly, to better qualify its workforce, and to put an entirely new generation through higher education.
After all of these adjustments, however, a general trend remains: The more complete intelligent computers become, the closer they get to achieving human general intelligence and the more they drop the limitation to individual fields, the more jobs they will be able to take on at reduced cost. And they have a special advantage as well: Intelligent computers are scalable. The software can be duplicated at a click of the mouse – a clear contrast to human experts. This makes machines that possess general human intelligence a low-cost alternative to human labor. This will simultaneously create a large amount of available “workers,” which in turn will cause wages to sink. And when these mechanical and easily copied workers are capable of doing virtually every job both better and cheaper than the human competition, what happens then?
After 2040: How will humans respond? We will become more intelligent ourselves!
For centuries, there has been an oft-used human recipe for reacting to increasing competition through machines: Humans attempt to optimize their minds and bodies so that they can remain competitive with the increasingly fast machines. Compare the daily work volumes and speeds of the average office worker today with the generation of their parents and grandparents, and you will be tempted to believe that the grandkids have themselves become computers. And perhaps, on some levels, this is true.
The boosting of human brain performance is one of the key phenomena of today’s post-industrial age. Through enhanced learning, intense multitasking, and drastically extended workdays, we are training our bodies and minds to compete with computers. Obviously – without realizing it! We call it normal.
You don’t need to be a prophet to foresee that the striving for the increased performance of the human brain as it exists today will continue far beyond the limits of present mental training. The trend is already in full swing now. Medications that increase concentration and memory power enjoy growing demand, despite the fact that the effects of current smart drugs are still extremely modest. The future successors of Ritalin and company will offer stronger effects with reduced side effects.
Added to this by 2040 will be the effects of the rapidly expanding field of genetics. Whether we like it or not, the selection of embryos promises an increase of over 100 IQ points per newborn in the coming years. This would mean that the average person would have an IQ equal to that of the most intelligence person living today. An even greater effect may be felt through the potential future creation of custom-tailored genomes.
One thing seems sure: Humanity will become more intelligent! Not only because countless people living in the world's poverty regions will be relieved of the curse of life-long intelligence impairment through iodine deficiency. A significant boost in intelligence is also expected in developed countries. The consequences of this “new normality”? Fewer learning disabilities. Significantly more university degrees. Higher incomes. Much greater numbers of highly intelligent individuals. It is estimated that there will be 10-10000x more “normal people” in the future who possess the IQs of today’s most gifted scientists.
The other side of this coin is that people will tend to become more similar. Why? Because genetic synthesis could allow us to free the genome of a given embryo from its accumulated mutations. The result would be future human beings who are less “deformed” through genetic mutations and who thus also become sick less often and are more intelligent, more resilient, and less varied in appearance. Would that be a good thing, or a bad thing? That judgment is left to the individual.
Of course, this situation would not arise in all countries on earth, because moral and religious concerns will oppose the selection and engineering of embryos just as in the past. For understandable reasons, Germany is also unlikely to be a pioneer of these advances. However, China and Singapore, for example, exhibit much more liberal politics. It may be that, in the future, they will not only permit new technologies for increasing the intelligence of their populations, but will explicitly facilitate their development.
And if countries like this are successful, then other states will hardly be able to refuse to do the same without the risk of ending on the sidelines of the world – along with their populations. If you project the progress and speed of today’s genetics research into the future, this would mean that the necessary technologies would be ready for mass use in 2035, and the first genetically enhanced generation would come of age in the year 2060.
Should we work to stop this? I don’t believe that would be reasonable. Because what is being described here is nothing more than the (albeit accelerated) continuation of natural evolution. For centuries now, evolution has led to the situation where the intelligence of new generations is often considerably higher than that of their parents and grandparents. Why should that not continue? Or, as Nick Bostrom says, “There are no grounds for assuming that homo sapiens is already the best possible cognitive biological system." In our current form of existence, we are best described not as the most intelligent species of all time, but as "the dumbest possible species capable of founding a civilization."6
After 2040: What will people do who no longer have jobs? Work!
Nick Bostrom, who with his book "Superintelligence" has animated leading thinkers like Bill Gates, Elon Musk, and Stephen Hawking in warning the public about intelligent computers, cites the history of horses in his prognosis on human labor potential.7 Horses were also once the most important means of production. Over the course of technological progress, they were supplemented with plows and carriages, which considerably enhanced their productivity (and here lies the parallel to human labor in relation to computers). As technology continued to develop, however, and plows became cars and tractors, the horse was gradually pushed aside. From a US count of 26 million horses in 1915, the beginning of the 1950s saw only 2 million.
There is a great temptation here to predict the same fate for human labor. But be careful! Following the drop to 2 million horses in the 1950s, the horse population in the US rose again to 10 million! Why? Did they get back their old jobs on the street and in the field? Of course not. However, economic growth made it possible for more and more Americans to own horses as a hobby. In other words, the horses got new jobs!
Human beings will no longer manufacture products, but identities
It is very probably that human labor will meet the same fate as that of the horse. Because when intelligent computers will become a preferable alternative for nearly all jobs by 2070, we humans will still have one area where we will not accept the work of computers, inexpensive and otherwise ideal though they might be.
You can already get a good feeling for this today if you look at the new "customer segmentation for the digital era" with which the 2b AHEAD ThinkTank tests the digital strategies and business models of its customers even today:
In the lower segment (economy), customers will always look for the best value for money. In this area, human jobs will disappear step-by-step.
Simultaneously, however, the upper segment (premium) will continue to grow. And here new jobs will appear for human workers.
The difference is: Humans will lose almost all jobs related to product manufacture, but they will gain masses of jobs in the manufacture of identities. Human workers will become identity carriers and identity managers. They will advise and support other humans. Instead of cars and houses, human laborers will provide recognition, belonging, and identity. In this context, doctors will become health coaches, insurance brokers will become risk coaches, teachers will become career coaches, food merchants will become nutrition coaches, cooks will become cooking coaches ... and so on.
Or, for those who prefer things more esoteric: "Following the conquest of the geosphere (physical world) and the biosphere (biological world), humanity is on the way to the noosphere. The goal of human labor and effort will be the conquest of consciousness, perhaps even a collective consciousness."8
But will we humans be able to live from "identity production"?
To put it bluntly: We are speaking about a vision of the future following the year 2040 in which we humans could increasingly lose the production and knowledge jobs that are normal today. We are talking about a time frame through 2070 in which the number of identity-creating jobs will rise sharply. These will be job descriptions that we do not yet know today. This also probably means jobs which we could not today imagine would be feasibly capable of producing income. Or can you imagine working as a "smiley"? The job description is simple: You walk through the city and cause people to smile. Basic smiles automatically earn a few dollars; hearty laughter and we're already talking twenty. And if you've really made someone happy, two hundred dollars will be automatically credited to your account.
From today's perspective, it is understandably difficult to imagine that everyone could have the ability to earn money in this, or similar, ways in the future. After all, where should all the money come from?
And this takes us to the economic side of things. Nick Bostrom calculates that for some years now, the share of global income consisting of capital interest amounts to roughly 30%, while 70% stems from wage earnings. In the age of superintelligent computers, the machines will have almost completely replaced human workers. The marginal costs of these computers will drop to virtually nothing, while the capital interest share in the gross global product will rise to nearly 100%. Simultaneously, the overall capital income will rise rapidly, because superintelligent labor would lead to constant technological advances and to new land ownership and raw materials through the colonization of outer space.
"If humans continue to remain capital owners, their income would reach dizzying heights, although no human being in this scenario would still have any significant wage income."9
Of course the precondition here is that, at the beginning of these unbelievable gains, all human beings would have at least a small positive starting capital. Otherwise this giant new wealth would simply pass them by. For all of our children today who at the beginning of this transition phase have no private wealth, or not even a bank account, or who have an account with a negative balance, this prospect is not exactly rosy. According to Bosrom, they would be forced to work as identity managers to earn a daily wage, or to hope for the aid of philanthropists who would distribute their unbelievable newfound wealth to the masses.
Unconditional basic income
The hope for philanthropic generosity and for one's own rise from being one of the dishwashers of the identity management market is naturally a very American perspective of life (although Bostrom was born in Sweden and currently lectures at venerable Oxford University in the UK). Europeans would probably feel more drawn to an alternative solution: The concept of an unconditional basic income, now promoted by various players for some time. Not only dm founder Götz Werner10 is propagating this redistribution mechanism in times of declining workloads. Even Rob Nail, CEO of Singularity University and perhaps the most avid and optimistic proponent of superhuman intelligence, universally proposes this form of performance-independent basic financing for all people.11
And even the chairman of a German DAX corporation believes that the digital world will take this direction. Telekom head Höttges suggests that, in times when value will be created more and more by machines and increasingly less through human labor, then earnings would also have to be taxed as mechanical labor and not human labor. These tax revenues could form the basis for an unconditional basic income. "It could mean a solution – not today, not tomorrow, but in a society that has been fundamentally transformed through digitalization."12
Even a few years ago it was easy to dismiss what was back then often a largely ideological debate about unconditional basic income as "typical German blather." But the discussion is much less "German" than we all think. Internationally it is even more intense. The magic word here is "universal basic income," and has even acquired the commonplace abbreviation UBI. The list of proponents reads almost like the Who's Who of international business. The Chief Scientist of Baidu Andrew Ng13 is on it, as is Chris Eliasmith14, one of the pioneers of artificial intelligence and Director of the Centre for Theoretical Neuroscience, as well as Nobel Prize winner Sir Christopher Pissarides.15
Even some individual nations are currently starting pilot projects in order to test the effects of unconditional basic income, such as Finland16 and the Netherlands.17 Preparatory studies exist, as elsewhere, in Canada18 and Switzerland.19
Jobs are for machines – life is for people (Scott Santens)20
There is, however, a prerequisite for the forward-looking optimism of UBI adherents. The essential precondition is that the world's population will not grow faster than its economic power. Nick Bostrom quite rightly points out the fact that most of human history has been dominated by the "Malthusian condition": More people were born than allowed for by economic potential. For this reason, most human beings have always lived at the edge of a minimal existence. "From time to time, wars, epidemics, or climate change decimated the human population. Then more space was available for the survivors. These were the times of economic miracles and growth ... until the gaps were filled again and the Malthusian condition was reinstated."21
The key question about the future for politicians is, therefore, not only that of an unconditional basic income, but also of the boosting of economic growth versus population growth. Globally! Because even if population growth is being kept low in some parts of the world, this is not happening in other places ... and this will lead to conflicts about the redistribution of wealth, at the end of which "everyone would possess the same amount – namely as good as nothing."
The algorithmic economic system
Nothing new?! True, the correlation between population- and economic growth is old hat. Only that the problem could really take off in a few years' time. Because in the age of the "algorithmic economic system," we are not necessarily only talking about human beings when we discuss the subject of population. What will happen when we no longer only distribute the world's available wealth among its human population, but when supercomputers start wanting their piece of the pie as well? Note that we are talking about computer programs that can be duplicated and cloned in milliseconds and are capable of occupying all of the world's available hardware in a minimal time. Which leaves us – much more quickly than we thought – back at the Malthusian condition again.
It could be the case that biological humans in this hypothetical future would eke out something of a niche existence, living among countless artificial intelligences and human-computer hybrids. The world around them is technologically advanced. There are not only intelligent machines, but also foods for biological humans that prevent aging, that boost intelligence and enhance beauty, that produce happiness.
For those of us in this scenario who cannot afford this, there are technologies that keep the costs of human existence low, such as foods that slow biological growth and metabolism in order to reduce the cost of living.
Some questions remain …
If you ask for my opinion: I would like to make a plea for calm here. In the next 20-30 years, we will neither experience superhuman intelligences nor masses of job-destroying computers. Because, until that time, we remain in the first stages of digitalization. In all honesty, no one today can make any guarantees about what will come after that. I consider the probability very great that humanity will make positive use of this technological development as well in the long run. So far it has always been that way.
But we also need to seriously consider the risks. In my estimation, these lie not so much in the loss of jobs. We are sure to find excellent solutions there.
The great risks lie rather in deeper questions of human freedom. For so far we have found no satisfying answer to the great questions about power in the future:
+ Who will control the computers with superhuman intelligence?
+ Can we impose upon uncontrollable computers a humanistic reason that would cause them to always act for the good of humanity?
+ And: How will conscious machines feel? How important is their well-being to us?
These are the questions for the third part of our series on artificial intelligence. The future experts of the 2b AHEAD ThinkTank will give you the answers shortly. Stay tuned!
We wish you a great future!
1 The comprehensive picture of the future by the 2b AHEAD ThinkTank on working environments in 2025 can be found in the book: 2025 – How We'll Work in the Future at: http://amzn.to/1plO2R1
2 These figures are based on the mean values of several surveys of the top 100 experts for artificial intelligence compiled by Nick Bostrom. See Nick Bostrom, Superintelligence, pp. 37 f.
6 Bostrom, N., Superintelligence, p. 69 f
7 Bostrom, N.: Superintelligence, pp. 226 ff
8 Kreisberg, J. C., in Wired 01/1995: A globe clothing itself with a brain; www.wired.com/1995/06/teilhard/
9 Bostrom, N.: Superintelligence, p. 228
11 Rob Nail, in: Die Welt; www.welt.de/wirtschaft/webwelt/article152198869/In-29-Jahren-sind-die-Probleme-der-Menschheit-geloest.html
12 Timotheus Höttges, in: Die ZEIT; www.zeit.de/2016/01/zukunftsvisionen-timotheus-hoettges-roboter-technik/komplettansicht
20 Scott Santens in: The Boston Globe: Robots will take your jobs; www.bostonglobe.com/ideas/2016/02/24/robots-will-take-your-job/5lXtKomQ7uQBEzTJOXT7YO/story.html
21 Bostrom, N.: Superintelligence, S. 203 ff