The future of Work
Perhaps the two most pressing concerns of Type 1 at this moment, are that AI will continue to bring significant job losses and contribute to increased income inequality. We'll look at these two issues in this article, because it is AI, not globalization or free trade or immigration, which has been the main cause for job losses. As we have seen in 2016, the anguish brought about by these job losses has been the main catalyst for a marked change in the political climate in the U.S..
Since the issue of job losses due to AI will certainly continue to heat up, it is worth placing it in the larger context of present day politics. Two political movements are vying for our attention currently and so they will affect our view of AI: populism and progressivism. Although not always the case, populism is currently on the right and progressivism on the left. Both movements have posited that "the system is rigged" by the establishment (which historically always seems to be the case!), although who inside the establishment is doing the rigging differs between them: for populists it is the corrupt political class who is doing the rigging, for the progressives it is the economic power class. We focus on populism, for two reasons. First, because there is currently a wave of populism affecting many countries, not just the U.S.. Secondly, because it places the blame for job losses on illegal immigration and globalization/free trade, which as we will see below are false causes.
It is not clear to most people right now, but job losses due to AI are far surpassing the losses due to either immigration or to globalization/free trade; some estimates put the proportion of job losses due to AI at 80% of all losses. AI may eventually compensate and add more jobs, but the nature of those jobs is unclear at this time. Because California is at the epicenter of all these (real or perceived) causes, it offers a good case in study. For those of us who live in the San Francisco Bay Area, the one thing we do not wish to happen, after we dreamed of a future brighter than ever because of technology, is for technology to become a sclerotic part of the establishment, become part of the "rigging", and draw the ire of the people who will loose their jobs because of it. Here come California's many dilemmas, within the context of the rise of worker dissatisfaction and the rise of populism:
Healthcare will likely be the one industry in the U.S. to be most affected by job losses due to AI. It is no secret that healthcare is a bit of a national embarrassment, the inefficiency of the system being evident to all those who come in contact with it, whether medical professionals or patients. In 2017 the costs were a staggering 10K per capita; you can see in the diagram below that the U.S. healthcare system is a bad outlier relative to all other countries; an outlier is a point that strays far from the main line, a bad outlier is an outlier above the line; Switzerland is the dot below U.S., at 8K and a higher GDP per capita; the far right is tiny Luxembourg, a country with a very high GDP per capita who still managed to have a per capita healthcare cost of just above 6K; Luxembourg is a good outlier. The lowest point on the chart, in the far left corner, is for Mexico.
Combine this inefficiency of the U.S. healthcare system with the strongest medical high-tech R&D in the world, and you have the perfect incentives for introducing AI at scale in the U.S. healthcare, which in turn will lead to significant job displacement. Since the healthcare industry employs 11% of all private-sector workers, the loss of employment in this industry will have substantial consequences for the rest of the economy.
How should people who would like to have careers in healthcare prepare for this outcome? Let's look at one of the most obvious examples. Because AI excels at recognizing patterns in data, it is able to read and diagnose radiology exams much better. Will we still have radiologists? Of course, but their job will be much different. Humans still need to understand how to program those AI systems and train them in order to recognize unusual patterns formed by disease. Recall from the background article Main Concepts that neural network training is more of a craft than a science. Data scientists develop a feel for what kind of layers they should use for different applications: convolutional layers for image recognition, LSTM layers for speech recognition, etc. The same kind of craftsmanship will be required of radiologists, who will continue to develop the medical skills for understanding patterns of disease, but apply those skills to train neural networks, instead of reading those radiology prints themselves. It is quite possible that medical training will require 12+ years of study (including years of training in data science) and produce a very small number of star practitioners, instead of the current level of medical school graduates. Even for those determined and lucky to practice, would that graduation mean hundreds of thousands of dollars in student loans with just a few years of practice to pay them back? These are haunting questions:
As we saw in the second video above, which was posted in September 2016, populism has been riding on the angst of economic insecurity, but without proposing bold policies to address that insecurity. Progressivism, as exemplified by Bernie Sanders' campaign, had also been riding that angst of economic insecurity, and it did propose some very specific policies to address them. These dynamics from 2016 have remained largely unchanged, and the economic insecurity is still continuing to rise. Since AI has been the main cause of this rising economic insecurity, we should perhaps try to diagnose better what happened in 2016, and understand why is it that neither populism nor progressivism are focusing on the real cause, even today.
The political center lost ground in 2016, to both populism and progressivism. The one candidate who exemplified competence, experience, and dedication, had been successfully boxed in and labeled corrupt, without any evidence. Despite not coming from privilege and despite a lifetime of hard work for child, family, and women causes, Hillary Clinton had been successfully tied to Wall Street and the establishment, through concerted efforts from both sides. Not easily inclined to entertain or charm her audience, she had been pegged to political correctness. In fact, attacking her had become the new PC and a bit of a national sport; the chants of "Lock her up" will stand in infamy in our history. We have been teaching our children that serious work and tears in the back room are less important than the giggle and the entertainment of the TV game in the front room.
In the presence of AI systems, listening in and many times producing data, the need to monitor truth in the data given to them and the data coming from them is essential. As we saw in the case of Facebook in the background article How We Form Political Opinions, AI-based messaging is targeted at our limbic system. In 2020 it will be coming at us from 3 directions again: our politicians, our social media and our foreign adversaries. But this time the stakes will be higher. So, let's review one more time:
The video is even more alarming because these kinds of sentiments are still present and they are coming at the very wrong time for the U.S.; the arguments about job losses due to globalization or free trade or immigration, as they were presented before the 2016 elections, are still clogging the airways. The difficult answers to the question of work in the age of AI must include a deeper analysis of human resilience, and different approaches to education, especially in those parts of the country most affected by job losses; we will elaborate on this later, but it is already clear that simplistic solutions (i.e., just have everyone learn how to write computer code!) will have to be revised and more robust legislative approaches will be needed. Education in the U.S., especially in the pre-college years, needs massive attention and strenuous thinking, because very little of what worked in the past will work in the future. The pain is real and it will get worse:
False assessments will increase divisiveness in our society, and they are coming from all directions. Trump supporters are not racist (only a tiny minority are), and many of them are tired of being told that they are. After 9/11 and the ISIS atrocities, they are justifiably concerned about Islamic terrorism. Trump and Sanders supporters are tired of the explanations given to them as to why their jobs went overseas (which as we already knew, it was not the case: the jobs went to the AI productivity gods, not overseas). Continuing to mis-diagnose the critical issues people in large swaths of the country are facing nowadays is a recipe for a potentially disastrous gridlock to come. The emptying of the political center, and the scattering of both Republicans and Democrats to the edges, does not bode us well in the coming age of AI, when competition with China will heat up, and jobs will vanish. The pain may be alleviated by learning how to program a computer, but it will not be the panacea to all problems. Are many people going to lose their economic value altogether, and face society irrelevance? What sociological and educational mechanisms should we set in place in order to avoid this irrelevance and allow dignity to continue? Here is a start:
The idea that AI will force us to become better humans is a fundamental one. Of course it will be essential that STEM (Science, Technology, Engineering, and Math) subjects be given special attention and careful evaluation, especially in early child development and proceeding all the way to high school. Computer Science will for sure have to make its way into earlier curricula. But these disciplines will not be sufficient to prepare us for the massive disruption coming ahead, the characteristics of which we do not know, and for which history does not offer us helpful clues. Humanities cannot be neglected, and we may have to redesign school curricula so that they will emphasize the fortification of the characters of our children, with real life stories of courage and overcoming of difficulties. One of the most intriguing aspects of our current predicament is that at this moment AI seems to have exactly the opposite effect:
There are deeper effects of social media on our well being. The constant and carefully curated information coming from friends, in which they appear to live fantastically wonderful lives, make many people feel inadequate. We had to invent a new term for this, FOMA (Fear Of Missing Out). Social media and its effects will be featured prominently in our articles, for many reasons, all related to AI; AI is the main technology behind the increased effectiveness of social media. Before we move on, let's look at the need to exercise critical thinking and develop a more rational evaluation of social media with the ideas about human resilience presented above. The call to quit social media altogether is a bit extreme, but the main points in the presentation are very useful to keep in mind:
AI And the Rising Income Inequality
Among the many problems that need our attention in the U.S. (which is the modern liberal democracy we are focusing on), one stands out, one which is very threatening and uniquely American: the extraordinary disparity between the incomes of the haves versus the have-nots, which problem will be greatly exacerbated by the rise of AI. It may be obvious to you already why AI will heighten the income disparity crisis, and we'll come back to it in detail soon, but for now, let's listen to some work coming from ... where else but Berkeley. (the film "Inequality for All" can still be watched on Netflix or Amazon Prime)
To re-emphasize the idea about a degree rather than fundamentals, we switch from Berkeley to Breibart News and Steve Bannon, quite a switch. The current populist fire springs up from a loss of economic security by too many people. This fire was not always directed at the Democratic Party, it started as an insurgency against the Republican establishment. We see that insurgency within the Republican party in the clip. We will analyze that idea at length in the , where our main point is that the weakening of the Republican party is in no one's interest in the U.S., because AI issues will need two strong and evenly matched parties to find solutions. The point here is that Steve Bannon shows a principled stand, however disagreeable it may sound, and explains well why he will continue to support populist policies from the outside, against the Republican establishment. But while his stand seems principled, the same mis-diagnosis of what needs to be done is present; he includes Silicon Valley in his targets, but for the wrong reasons, as bastions of liberalism, which they are not, promoting a confrontation between government and Silicon Valley instead of working together to produce effective regulatory oversight in support of a beneficial development of AI.
Job losses due to AI will obviously impact the balancing needle between taxation and welfare. The Universal Basic Income proposal will obviously require higher taxation. The argument that is often made against welfare is that high welfare benefits will dis-incentivise unemployed people from looking for work. That view does not seem to reflect the core of human nature. In the Nordic countries, despite having high welfare benefits, people are quite serious about finding employment. What is the bedrock nature of humans, material comfort or dignity and finding significance in their lives through work? If we adopt a dim view of human nature, then we stand no chance in facing off the challenges of living in an AI world. We will refer to Aldous Huxley's "Brave New World" and the value of mindless pursuit of comfort/self-interest in one of the following articles, but let's view this issue in contemporary U.S. society rather than a dystopian futuristic novel:
A market-driven economy should be predicated on informed consumers who buy quality products and ditch the poor ones. A market-driven and AI-dominated economy will require even more informed consumers. The advertising industry already makes heavy uses of AI to influence buying patterns, and that trend in the usage of AI will accelerate. And just in case you are ready to dismiss Chomsky's argument that people are not "homo economicus" (i.e., just selfish pursuers of comfort), simply because Chomsky is such a symbol of left-wing politics in the U.S., let's hear essentially the same argument from a self-made billionaire:
What is our message here, by including these video clips? It is not to criticize success, it is not to criticize the top 1% of earners, and especially the billionaires' club. They played by the rules of the game and won, which is admirable. It is the game itself that needs to be changed, and AI will force us to change it. It's not clear what a preoccupation with the billionaires will bring, instead of a preoccupation with the more constructive issue of how to change the rules of the game. The following quip is attributed to Larry Ellison: "Tell me what I can do with $3 billion that I cannot do with $1 billion". The relevance of money fades past a certain level anyway, and the human condition has no way to enjoy those irrelevant levels. There is no doubt that income inequality is a pressing issue in the U.S., and AI will greatly exacerbate this inequality. So the debate around wealth redistribution must be constructive, not destructive, and it must include AI matters. And that constructive approach means that some rules of the game must be changed so that the wealth inequality reaches a more benign curve which we can all digest better. The billionaires, many of whom made their fortunes in high-tech, seem to understand AI much better than the government, and many of them are actually leading the charge for a constructive revision of the rules of the game. As the speaker just said "Markets are not jungles, they are gardens; they need to be tended."
Will U.S. continue to be a land of opportunity or is that opportunity becoming unequally reachable, and dangerously so with the rise of AI? Steve Bannon is preoccupied by the same question as the "liberal Californians" in his crosshairs, so the good intentions on both sides should be a reason for hope. The seriousness of the problem of AI exacerbating income inequality weighs on anyone interested in our future prosperity.