Smart, quasi-autonomous robots and machines are replacing humans in workplaces all over the world. They learn fast, work hard and complain less. Intelligent technologies are increasingly delivering greater value for less money.
But “better than human” comes with its own managerial challenges. What happens when these algorithmic ensembles underperform? Who retrains “machine learning” underachievers? When sophisticated robots — Fidelity robo-advisers, Uber autonomous cars — behave in ways that make customers nervous, how do they get the feedback essential to improvement? Who — or what — is accountable?
Brilliant, hardworking machines will require job reviews just as humans do. Executives who can’t get their robots to do a better job may lose their own. Empowering smart machines to — pun intended — live up to their potential may well become the essential new 21st-century leadership skill.
Increasingly, robots and machine learning systems will be held to the same key performance indicators that govern human accountability and effectiveness. If the automated chatbot isn’t boosting the contact center’s customer satisfaction rating, reprogramming or retraining is required. Should the manager look to the company’s best customer care representatives to fix the problem or should he look to the technologists? Should robo-advisers be working more closely with their human counterparts?
The same leadership, managerial, and motivational questions that haunt smart human organizations now stalk today’s globally networked cyborganziations: Would 360-degree performance reviews lead to new insights and efficiencies? How should smart machines be programmed — or trained? Are there new networks, such as the “Internets of Things,” or novel data sets managers should access to assure ongoing performance improvement?
Organizations managing smart robots and high-performance machine learning ensembles must monitor correlations between algorithmic efficiencies and customer value. The challenge will be deciding how much those correlations are defined and determined by human leaders as opposed to artificially intelligence support systems.
“We simply don’t have good computational theories for social norms and behaviors,” observes Jerry Kaplan, a Silicon Valley serial entrepreneur and author of “Humans Need Not Apply.” That makes it difficult for smart people and smarter machines to productively coexist and collaborate. But there should be no doubt that people running the business are going to use machine-generated data and predictive analytics to make the machines that generated them even better and smarter.
The most provocative question going forward is whether tomorrow’s organizations get better results and more value from performance reviews of their best people or their best machines.
Michael Schrage
(Michael Schrage is a research fellow at the Sloan School’s Center for Digital Business at the Massachusetts Institute of Technology. His latest book is “The Innovator’s Hypothesis.”)
