It was a packed room over on the Chicago Booth London campus on Wednesday night as alumni and the business community came together to hear Prof Mike Gibbs give a masterclass lecture on his thoughts on the potential labor market robot and AI apocalypse. Mike wrote the widely used textbook, Personnel Economics in Practice, which gives a framework for understanding organizational design and managing employees, so his views on AI taking over jobs is quite helpful. We ended up having a lively and interactive discussion on the dramatic effects that technology has made in the past on jobs and productivity and what to expect in the future.
The first approach at evaluating job losses to robots and AI is to look at the type of jobs at risk. According to Mike, jobs are designed based on four characteristics: degree of interdependence, level of multitasking vs specialisation, level of discretion given to an employee, and finally the level of employee skills required. At one end of the spectrum, there are “classical” jobs which rate low across the four characteristics. These are jobs that focus on efficiency, uniformity and control (i.e. factory line workers) and as we have seen, lend themselves to automation. On the other side of the spectrum are “modern” jobs, which rate highly across the four characteristics. Here, the worker is continually learning and the tasks are more complex, evolving or unpredictable. In these environments, “modern” jobs are less likely to be automated in the foreseeable future.
The other approach examines the aspects of a person’s ability that defines their job. Mike referenced the work done by Frey and Osborne from the University of Oxford. In their work, they noted that there were three key bottlenecks to AI taking over our jobs:
1. Perception and manipulation
A robot is unable to take a phone from someone in contrast to Mike snatching my phone as I was busy typing notes from his talk. Physical ability hasn’t been automated yet (although just look at the Boston Dynamics robot videos or the videos of robots pouring beer to see how that’s evolving).
2. Social skills are difficult to automate
The mishap with Microsoft’s Tay chatbot shows that we’re still a long way to go. I also don’t see Alexa turning into a ‘Samantha’ from the movie, Her, anytime soon.
3. Creative intelligence
Many jobs still require originality of thought and the ability to develop creative ways to solve a problem.
Within their work, Frey and Osborne looked at 700 job occupations and concluded roughly 50% could be automated. A scary result, but one that’s not surprising when looked at in the context of Mike’s classical vs modern framework. Some of those occupations included sales and service roles. One has only to look at how CRM software and computerised call-centers have affected jobs.
Should we be concerned? I actually took some comfort from one major labor disruptor in mankind’s history. Mike shared with us a chart looking at the introduction of women into the workforce. It was a big disruptor at the time. Women initially took over men’s responsibilities as men were off at war; however, after the war, we did not see a severe job loss for men or see women go back to staying at home. For me, I’d like to believe that AI and robots will result in the same outcome. Technology throughout history has shown opposing effects: in some cases substituting humans, but also other times complements and augmenting humans.
Finally, for the next generation, Mike noted that in addition to teaching STEM – science, technology, engineering and maths – at schools, he believed adding more focus on economics (i.e. using utility curves to determine AI decision making) and statistics will further ensure more modern jobs are created (STEEMS).
Some more reading of Mike’s work can be found here.