As A.I. Takes Off, We Need a Plan to Deal with Societal Disruption
As an Xconomist, I have been asked to predict what technology might really take off in 2018. My response is something of a “cheat” in that it relates to artificial intelligence (AI), a technology that has been poised to take off next year for the past several decades and a topic about which I have not written in decades. Way back when computers had (single digit) megabytes of memory, AI was all the buzz. I worked in the field and among my colleagues were many of the founders of the field, several of whom made amazing claims about AI—specifically, how it would change the world.
At that time, we were wrong. AI was over-hyped and what was delivered failed to meet expectations. Consequently, AI actually became something of a taboo—it was not mentioned in polite circles for years.
With advances in technology, however, AI is making a comeback. But first, let me put things in perspective: today, an Intel Core i7 6950X, a microprocessor that costs less than $2,000, has more than 300,000 times the processing power of the DEC VAX 11/780, which was the mainstay of much of the AI research in the 1980s—and that cost more than $300,000 in today’s money. (In other words, computing power per dollar today is about 50 million times better than it was just 30 years ago.)
We have seen computers tackling problems that were once believed to require human intelligence. Chess-playing programs are now capable of regularly beating the world’s greatest players, and a computer recently beat the world champion Go player. While it is quite true that these problem domains are very limited in scope, the fact that just 20 years ago, the best-in-class chess programs were very poor players whereas today’s best-in-class are the world’s best, shows the remarkable progress that has been made in AI.
Closer to home are the advances being made in automated driving. This technology has advanced to the point where there are fully automated “taxis” in operation in Pittsburgh (a fitting locale, since the technology was originally developed at Carnegie Mellon University decades ago). I predict that while still in the experimental phase, it will not be long (less than 10 years) before driverless car become the norm.
However, what is far more interesting than the technology itself (and the technology is downright fascinating) is how we, as a society, are going to deal with it. The societal changes that are likely to occur as the technology becomes part of our everyday lives over the course of the next one or two decades will make the previous “revolutions,” e.g., the industrial revolution, look like minor uprisings by comparison.
Let’s stay focused on cars for a minute. From a technological perspective, automating long-haul trucks is quite a bit simpler than in-city taxis. In fact, platooning (where a lead vehicle is driven by a human driver with a number of self-driving trucks following the lead) exists today; and fully autonomous trucks can probably be deployed within the next few years.
At the potential cost of 3.5 million jobs.
And this is just the tip of the iceberg.
When autonomous driving can be used for in-city transportation, a further quarter-million taxi drivers stand to lose their jobs, as do 700,000 bus drivers and 500,000 ride-sharing drivers. The impacts do not stop here. The need for parking will be greatly diminished, again putting more people out of work and cutting off a revenue stream upon which many municipalities depend. More interestingly, the need for cars will be greatly diminished (most cars are parked for 95 percent of the time), which will lead to massive layoffs in the auto and related industries.
As important as cars are to today’s society, they are just one small aspect of it. Last year, IBM Watson took all of 10 minutes to correctly diagnose a patient whose specific type of cancer had stumped numerous doctors for months. It is also generally believed that the days of human financial advisors are severely limited because AI-driven programs generally provide better returns than do humans. Even the task of writing, one that seemingly requires a human touch, is being automated, especially for factual reporting (such as sports recaps, corporate earnings reports, and recently a chapter in a best-selling fictional series). This is at least another million-plus jobs at risk—decent-paying, prestigious jobs.
But, my estimate may fall far short of the mark—no less an authority than Merrill Lynch recently stated that “47 percent of U.S. jobs have the potential to be automated.” The financial planning firm also said it anticipates “$14-$33 trillion in annual creative disruption impact in 10 years [which is approximately 25 percent of the global GDP], including $8-$9 trillion in cost reductions across manufacturing and healthcare, $9 trillion in cuts to employment costs via AI-enabled automation of knowledge work and $1.9 trillion in efficiency gains via autonomous cars and drones.”
While in previous technological revolutions, it was generally believed that those who experienced displacement were those without more advanced knowledge/training/skills, the AI revolution will cut across socioeconomic lines like nothing has done so before. Both highly trained professionals and minimally skilled workers are equally at risk to lose their jobs.
Which gets us back to the question of how will we deal with this from a societal perspective.
The first thing to note is that a) no technology since the beginning of time is inherently good or evil, and b) that technology, once invented, cannot be un-invented. We, as a society, need to make informed decisions about how we are going to deploy AI technology. The right decisions may lead to a prosperous future, such as that depicted in Star Trek; whereas the wrong decisions would likely lead to a dystopian future, perhaps one as bleak as depicted in Terminator.
These decisions are extraordinarily difficult to make. For many people, especially professionals, their self-identity is entwined with their career and how much money they make. What happens if they no longer have a career? What can large segments of the population do if there is simply no need for human labor? It has been suggested that providing everyone with a minimum livable wage and encouraging volunteerism is an alternative to labor—but is it? Such an approach will likely exacerbate wealth disparities, and historically, this has led to significant societal unrest.
Second, whereas many have touted the importance of a solid STEM (Science, Technology, Engineering, and Math) education, there is an increasingly important need for leaders to not only understand STEM, but also liberal arts. Engineers are more than capable of building anything asked of them; however, they must also have the training to be able to ask why and consider the potential long-term consequences.
Third, the national discourse must focus on the real issues that threaten to upset our way of life. The number of workers who will be displaced by the rise of the robots are countless and will not be stopped by any type of physical wall. The current “debates” about wall-building and limiting immigration are no more meaningful that rearranging the deck chairs on the Titanic—we must get our leaders to address this critically important matter before it is too late. (And we have about a decade before it will be too late.)
In summary, 2018 will be the year of AI. Or, perhaps 2019 will be. Those predicting the rise of AI have been wrong countless times before and I doubt that I am any wiser than those who have made such predictions in the past. However, I can make two statements with certainty:
1 – In the near future, say within ten years, AI will play an important, and possibly dominant, role in our society and around the world.
2 – Unless we start taking action today, we will be unprepared for the rise of AI, and without proper preparation, the consequences will likely be undesirable for large segments of society.