Dr. Peter Bishop
This new magazine, Age of Robots, contains a clear message: we are entering the age of robots, and we should learn now about what this age might be like. But how do we do that? Do we simply extrapolate from the age of machines? We could, and many do, but robots are not just machines. They are, or will soon be, intelligent in many ways beyond machines. Should we extrapolate from the age of computers? Again we could, but while the core of a robot is a computer, it is much more. So how are we to learn and in some sense know what is ahead of us in this age of robots. Read this article to find out.
Prediction is difficult. In fact, it is not only difficult but fundamentally impossible for many different reasons.
Then why do we do it? One reason is that we dislike uncertainty. We want one definitive answer to every question and one best solution for every problem. That’s what we learned in school—you get to raise your hand when you know the answer, and the right answer gets you an A. The search for the answer and the solution is drilled into us from the moment we first step into a classroom. It’s no wonder that we want prediction—giving us nice clean, simple, definitive predictions, even when they are often wrong.
The alternative is to ask open-ended questions that have multiple answers, such as: How will more robots affect people’s work in the future? Unfortunately, most people take that as a simple, one-answer question. Ray Kurzweil, the new Director of Engineering at Google, has his answer: “The future offers meaningful work, not meaningless jobs.” Martin Ford, author of Rise of the Robots: Technology and the Threat of a Jobless Future (2015, Basic Books), has another answer: “Eventually we’ll get to the point where there won’t be enough jobs for most people.” Who’s right? Believe it or not, both are, because the future is not singular—one question, one answer; it’s plural—one question, many answers.
That’s the root principle of the emerging field of futures studies: the future contains many futures and many plausible scenarios, any of which could happen. Some futures may even have excellent empirical support. The present may be singular, but the future remains plural before it becomes the present. So let’s stop the argument—there is a future with jobs and a future without jobs. Both scenarios are possible and we need take both of them seriously.
We all need to give up the idea of a single “right” future and embrace the uncertainty of many plausible futures. Uncertainty is our friend. While it makes knowing the future difficult, it makes influencing the future possible. The future is not deterministic; it is not a mathematical equation where the result is already certain before the calculation is made. Fortunately, the real future is one we can influence since it’s not yet determined to be just one thing. Time is our friend.
The Traditional Approach to the Future
Seventh grade science taught us a basic approach for making predictions about the future. Take a theory, express it as a formula, note the initial conditions, and extrapolate the outcome to the chosen timeframe. This works well for simple systems or datasets.
When we don’t have a formula, we use the next best thing, our minds. We argue by reasoning, or perhaps just by intuition, on what we think the future will be. Philip Tetlock and Dan Gardner wrote about people who are good at prediction in their book, Superforecasting: The Art and Science of Prediction (2015, Broadway Books). It’s a great book, but it’s asking the wrong question. The question is not who is good at saying what the future will be (a single prediction), but rather how are we all to know what it might be (a multi-valued forecast).
The Futures Studies Approach
Forecasting using the approach from futures studies is a simple process that uses evidence to make inferences about the future. The steps in that process are as follows:
- Select and describe the domain. In this article, the domain is the future of robots in the United States in 2040.
- Gather information on the past and present conditions of that domain.
- Forecast the expected (most likely) future of the domain using evidence that points to where we are headed in the future.
- Identify and challenge assumptions required by that forecast. Every successful challenge automatically creates an alternative future.
- For instance, trends might continue, but there might be a good reason why they might not; and if they don’t, we get a different future.
Look for (weak) signals of change. Every weak signal could grow to be a strong signal and shape the future in a way different from what we expect. (All strong signals, like trends, were weak signals at one time.)
- Describe a few of the most interesting and important alternative futures that arise from challenging assumptions and finding weak signals.
- Identify the implications of the expected and the alternative futures for oneself and for one’s family, enterprise and community, and/or for the world as a whole.
At this point we are ready to begin planning how to influence the future, to bend the trajectory away from the futures we do not want and toward those we do—the preferable future.
The Future of Robots
So let’s apply that approach to the future of robots.
1. Select and describe the domain
We will focus on the future of robots in the US in 20–25 years from now (ca. 2040). The key question is: “Will robots create more jobs than they eliminate by about 2040?”
2. Gather information on the present conditions and on trends changing the domain
- Between 1.5 and 1.75 million industrial robots work in manufacturing in the U.S. workforce—that’s about one robot for every 8 workers out of the 12 million who work in manufacturing.
- The Pew Research Center interviewed over 1,800 experts on robotics on whether robots would create more jobs than they eliminated. The group split just about down the middle (52% vs. 48%). So we have two plausible scenarios, each supported by qualified experts—the answer is not singular; it’s plural.
- Since the 1870s, technology has created more jobs in new sectors than it destroyed in the target sector (manufacturing).
- Technology has shifted work away from manual labor to services and knowledge work.
- The use of robots is growing at about 11% per year, according to Henrik Christensen, director of the Contextual Robotics Institute at the University of California, San Diego, and the number is doubling every 7 years or so to around 18 million by 2040. That’s more than the total number of workers in U.S. manufacturing today, so robots could well replace most workers by then.
3. Forecast the expected (most likely) future of the domain
Robots will clearly be a significant part of the U.S. workforce in 2040. If industrial robots are any indication, their growth rate puts them on a path to replace most manufacturing workers by 2040. Of course, some human workers will need to remain to tend to the robots. The result is likely to be an increase in productivity, which could mean less expensive products (as has happened before), and a lower cost of living.
4. Identify and challenge assumptions required by that forecast
Four (of the many) assumptions required by this forecast are:
- The trends listed will continue until 2040.
- The future is like the past, where increased productivity due to automation freed people from work in one sector and allowed them to find work in new sectors. So the assumption is that the future will be like the past in that new sectors of work will be developed.
- The increased productivity also reduced the cost of living so people could afford goods and services that they could not before. So again, the assumption is that the cost of living will go down this time as well.
- People accept the move away from traditional work and allow the transition to take place.
Every assumption has an alternative assumption that is literally its opposite. If the original assumption is that a trend will continue, its alternative is the opposite, that is, it will not continue. We can always say this, because the opposite is always possible. But do we have any reason to believe the alternative might come true? That’s the difference between an alternative assumption that is merely possible (it could happen) to one that is plausible (here’s a reason it might happen).
Here are a few plausible alternative assumptions:
1a. Alternative assumption: The trends listed will not continue to 2040.
Reasons that the alternative assumption might come true: These trends assume continued economic and technological growth, but many people have forecast dire consequences for climate change and/or resource scarcity that will turn the world’s attention away from technological growth to solving environmental problems.
1b. Alternative: The trends listed will not continue to 2040.
Reasons: The Internet could become so risky and unstable because of hackers or even cyberwar that the development of electronic technology comes to a halt.
2. Alternative: No new sectors of work open up.
Reasons: Optimists on the ability to find work point out that it was hard for people living in an agricultural economy in the 19th century to imagine what work would be like in an industrial economy in the 20th century. It is similarly hard for us to imagine what the new work sectors will be in the age of robots. After all, robots are now knocking at the door of service jobs, like truck driving for example, and more intellectually demanding office work. The most likely sectors that will be immune to automation are some form of design, creative or freelance entrepreneurial work. This forces the question: Can a workforce of 150 million workers survive on design and freelance work?
3. Alternative: People might not have the money to consume the goods and services in the new sectors no matter how cheap they are.
Reasons: If robots really do displace a significant portion of the workforce, those workers will be hard-pressed to purchase life’s essentials, much less goods and services from new sectors.
4. Alternative: People do not accept the move away from traditional work and block the transition.
Reasons: The Pew study cited earlier found three areas of significant concern:
Impacts from automation have thus far impacted mostly blue-collar employment; the coming wave of innovation threatens to upend white-collar work as well.
Certain highly skilled workers will succeed wildly in this new environment, but far more may be displaced into lower paying service industry jobs at best, or permanent unemployment at worst.
Our educational system is not adequately preparing us for work of the future, and our political and economic institutions are poorly equipped to handle these hard choices.
Any of these concerns could grow to become the overriding issue of the next era.
Now people can discuss the evidence and the assumptions required to use that evidence in an effort to reach an understanding of (if not agreement on) the various arguments for the forecast.
5. Describe the alternative futures that come from each alternative assumption
1a. Meltdown: Internet security declines, and crime grows to the point where people are reluctant to use the Internet for anything other than the most trivial uses. They turn to old ways of communicating, doing business, and entertaining themselves.
1b. The end of the market: The scramble for basic resources such as food and water becomes so intense that people disconnect from the market, preferring to go it alone.
2. The end of work: Robots take most of the traditional jobs, and few new sectors open up to give people work. The consequences could either be an era of leisure where robots are so productive that goods are so cheap that there is enough for everyone to support themselves on relatively small incomes, or they could lead to mass unemployment.
3. The End of Income: New sectors do open up, but they do not produce the incomes that people are used to in the traditional sectors. It’s hard to make ends meet in this future.
4. Backlash: People need to work to support themselves and to give them some meaning and purpose, so they pressure government to slow the rate of growth of robots.
6. Identify the implications of the expected and the alternative futures for oneself and for one’s family, enterprise, community, and/or for the world as a whole.
The basic implication is that we must prepare for alternative future conditions:
- We may find ourselves in direct competition with robots as they take over significant portions of work. Consequently, we need to identify the talents and skills that robots are unlikely to replicate in our lifetime and strive to cultivate them.
- Alternatively, we may need to collaborate with robots to help us do the work we have always done in better and more efficient ways. This future requires that we must learn to collaborate with machines more than we had in the past.
- Alternatively again, we need to keep an eye out for sudden disruptive change, both environmental and economic—the kind that could challenge our most basic assumptions about how we live and work. Surprising disruptions are unpredictable by definition, but we can prepare ourselves for sudden change by thinking about, or simulating, such changes in our thoughts and discussions, just as astronauts do when they train to live on the Space Station. While it might not be a pleasant future to contemplate, it could result in a less stressful and more meaningful way to live.
- Alternatively again and again, the degree of change might be less than we expect, leaving the world much more like it is today than we expected. So let’s not abandon the ways of living that have been successful for centuries, at least not just yet.
The future is really a set of multiple futures, not just the one future that most writers and thinkers propose, and even these scenarios are probably too definite to be real. The real future will be complicated, involving some combination of many of these scenarios. It’s rarely one thing, even when it becomes the present.
The Debate Rages On
Will robots take our jobs? Will we be able to support ourselves? Will we be able to feed our children? Will the machines be a boon or a curse in a future society?
Put the debate away! We do not know which future will occur, and we will not until it is upon us. Keep your eyes and minds open. Hold on to your vision and values, but recognize change when it occurs. It usually means that we must change the way we live those values, not a change in the values themselves.
And finally prepare today to live in three or four different futures. Learn how the world is changing and challenge yourself to change with it.