Robots did not overtake the world. But what about artificial intelligence?

Labor markets have been upended over the centuries by technological change.  Only John Henry could compete with the steam-powered drill, and he paid dearly.  In the not too distant past, there have been concerns about the impact of software and robots on wages and employment.  Now the big worry is artificial intelligence.

Stanford economics PhD student Michael Webb has developed a framework that can evaluate the impact of new technologies on different occupations by matching words between patents and job descriptions.  Looking at data between 1980 and 2010, he found wages and employment declined the most in jobs that were most exposed to competition from software (broadcast equipment operators) and robots (forklift drivers).  These are jobs that do not require a college degree.

Based upon patents related to AI, Webb finds that the occupations most likely to be endangered include clinical laboratory technicians, chemical engineers, and optometrists.  All of these jobs require post-secondary education.  Overall, the risk of displacement from AI is greatest for those with college and graduate degrees.

Technical change and globalization have resulted in the loss of a number of middle class jobs.  AI could be coming after the upper income brackets.

Machine Intelligence

Twenty years ago the internet was supposed to "change everything" and defy conventional economic analysis.  Nope.  Did not happen.  The internet lowered the cost of search, information, and communication.  New products spawned by the internet were characterized by massive fixed costs and negligible variable costs.  Key insights about the internet continue to follow from basic economics.

Now the hype is about machine intelligence.  Three faculty members at the Rotman School of Management at the University of Toronto have a short article on the HBR website where they argue that the economics of machine intelligence can be summed up as "lower costs of prediction."  This means that firms will have lower costs associated with demand forecasting and inventory management, leading to wider adoption of these practices.

As prediction becomes cheaper, there will be an impact on other inputs into the production process, depending on whether they are substitutes or complements for prediction.  For instance economists who make predictions may be displaced by machines.  The authors think that judgment skills will become more important, serving as a complement to cheaper predictions.  I am not sure what they mean by judgment skills, but presumably they are referring to cognitive processes where humans will continue to have an advantage over machines.

Machine intelligence will soon be coming to higher education.  Some business schools are already experimenting with using tools based on machine intelligence to drill newly admitted students on basic skills in math and statistics.  Will a machine-based socratic dialogue be next?