You may lose your white-collar job to a robot sooner than you think. According to the IDC (International Data Corporation), a global provider of market intelligence and analysis, big changes in the U.S. workforce will happen quickly and at a scale you would never have expected. The IDC’s report, “IDC FutureScape: Worldwide Robotics 2017 Predictions” outlines several jarring projections, one of which is that by 2019, a hefty 35% of industries in health, commercial goods and utilities will use robotic automation in their operations. Nearly 40% of robots currently in use will be connected to a cloud network, increasing their efficiency and ability to access vast quantities of information.
Not only will robots continue to play larger roles in manufacturing and service industries, but they will press forward in white-collar fields as well. Martin Ford talks about the implications of this scenario in his book, “Rise of the Robots.”1
He gives us an interesting preview of where the growth and already widespread use of robotic technology may take us as a society. In a gripping narrative which explores how technology has impacted various sectors of the economy to date, from manufacturing to retail and medicine, the author sketches out the effects of a greater robotic presence for the US workforce and the larger economy.
A Story of S-Curves: The Explosive Growth of IT
Moore’s Law refers to an observation made by Gordon Moore of Intel Corporation about the speed at which the number of transistors per square inch on an integrated circuit grows every two years. Moore claimed the number doubles every year, a prediction he said would continue until the year 2020. Ford shows how Moore’s law has been validated by the exponential pace of technological progress.
New technology has consistently followed a S-curved trajectory—exponential gains are followed by a plateau in new developments. This plateau in progress is the result of reaching certain limits in key physical laws. After a certain point, for example, transistors simply cannot get much smaller.2
What is unique about the field of information technology, however, is that one S curve representing steep and significant development often launches another S-curve. For example, Ford predicts that the potential is great to improve computer speeds even more by experimenting with networks of processors or by using carbon-based elements to rework chip design.
From Years to Split-Second Computation
Consider that in the early 1980’s, the state of computer hardware was such that a particularly difficult computing problem would require roughly 100 years to compute. Today, processing speeds have advanced to the point that a complex computation can be performed in a minute or less, representing a rate of increase in processing speed of no less than 43 million.3
Ford argues that information technology is like no other technology in recent modern history. How does this progress translate to the tools we are likely to see in the workplace?
What Does an IT Dependent Society Look Like?
Signs of an IT driven culture are everywhere. Electronic medical records are pulled up during doctor’s visits, banking processes are computerized. Our daily lives are dependent on data we provide our personal computers or smart phones. We depend on GPS units that spit out anticipated time of travel given traffic and road conditions. We ask Apple’s “Siri” questions and she tells us precisely what we want to know within seconds.
Ford calls IT a true “general purpose technology.” We have grown to depend on computers to perform all kinds of routine tasks. The increasing numbers of complex algorithms, the ability of computers to find new relationships given set inputs (aka machine learning), coupled with access to vast quantities of data available in the cloud means we are on the cusp of a time when computers can perform truly sophisticated tasks. Ford introduces the next part of his argument with unsettling predictions: “Computers will cease to be tools that enhance their [human] productivity and instead become viable substitutes (emphasis added.)”4 How does the replacement of a human workforce by a largely robotic one come to pass?
A pivotal moment in recognizing what computers were capable of came to pass when “Watson” defeated Jeopardy champions Ken Jennings and Brad Rutter in 2011. Drawing from a vast amount of information collected from everything from reference books to newspaper clippings, web pages and other material, a team of researchers built a computer that could take apart a cryptic clue, analyze the question and using thousands of algorithms, dip into the stored categories of data to sort and identify the best options. Watson had the ability to rank data and identify which algorithms were likely to produce successful answers, ultimately giving the winning response with nearly 100% confidence.5
IBM’s creators then went on to apply Watson’s cognitive skills to other fields like medicine. Once Watson was connected to the cloud, accessing reams of medical data in the form of journal articles, case studies and physicians’ notes, Watson could come up with surprisingly accurate medical diagnoses. Given a big enough data set, Watson’s creators concluded, identifying correlations between data sets became easy. This is what makes access to big data so powerful.
Combined with the revolutionary cognitive computational ability Watson represents, artificial intelligence (AI) moves to an entirely new level. Now Watson is being used to train oncologists to better deliver medical treatments. Ford points to multiple applications of machine learning already at work, Google translator for example, and autonomous cars.
Because humans could never be able to make sense of huge amounts of data, often the conclusions and work product computers produce exceeds in quality that which could be produced by a human.6 Genetic programming or machine learning combined with access to the vast data sets of the cloud represents a powerful combination.7
AI and the Engineer
Computers have composed symphonies, discovered new designs for commercial products, outcompeted humans in challenging contests like chess and Jeopardy and made strides in other complex fields. While Ford concedes that at this point computers are mainly performing routine tasks, the frontline is changing rapidly.
He claims that the white-collar jobs that are being offshored now are the ones that will be fully automated in the relatively near future.8 He foresees robots doing legal analysis and other tasks once performed by executives with years of experience. It is no secret that many of the world’s most respected thinkers in the field of AI are alarmed. Elon Musk has been a vocal advocate of proceeding with caution, not just because of the threat to jobs in the U.S., but because of machine-learning technology and the potential to put AI in a position of power relative to the humans that created it.
Could a robot with sophisticated cognitive abilities replace an engineer? I admit that I foresee our GeoHECRAS software automating more of the routine tasks of the civil engineer. Once raw data is entered, the software can perform much of the analysis that the engineer has typically done manually. Others seem to agree with this point of view. Computers will probably replace the mundane tasks of engineers fresh out of college.
But I see computers and engineers working as a team, not one taking responsibility from the other. I am skeptical of robots completely automating the work of seasoned engineering professionals. Is it likely that a computer equipped with AI could completely replicate the knowledge-base of a senior engineer? Could a robot equipped with AI arrive at an analysis that intelligently processes the many variables and considerations a complex engineering project usually involves? The question of world conquest by robots aside, I think the work of most engineers is safe for now. But ask me again six months from now and my opinion may have changed.
1 Martin Ford, Rise of the Robots: Technology and the Threat of a Jobless Future. (New York, New York: Basic Books, 2015).
2 Ford, Robots, page 70. For example, Brown predicts that by 2020 we will have reached the limits in terms of how small computer chips can be, in this case about five nanometers.
3 Ford, Robots, page 71.
4 Ford, Robots, page 118.
5 Ford, Robots, page 101.
6 Ford, Robots, page 91.
7 Ford, Robots, page 94.
8 Ford, Robots, page 112.