BOSTON—Concerns about the impacts of technology on jobs are nothing new. Fears about automation replacing assembly line workers, for example, have been around almost as long as technology itself, and to a large extent, those fears have been realized. The advent of big data raises concerns that technology might be threatening the jobs of knowledge workers. But can automation have the same impact on a workforce that trades in knowledge and creativity rather than physical skill?
Tom Davenport, professor at Babson College, Research Fellow at the MIT Center for Digital Business, and a Senior Advisor at Deloitte Analytics, addressed this question today in a keynote address at an event hosted by Massachusetts Technology Leadership Council titled, “Big Data and the Knowledge Worker: Impacts on Workforce and the Economy.”
“(Knowledge workers) have had a pretty good run over the past few decades, said Davenport. “It’s been pretty tough for factory workers, before that farm workers. It’s been pretty tough for service and transactional workers. Now that same automation is coming close to home. We always thought that whenever technology took over a type of job, that humans just moved to higher ground. When farming started to go away, people moved into factories. When factories started to go away, people moved into cities and did service and knowledge-oriented work. But this time, there is no higher ground.”
Davenport said automation has progressed from manual labor to administrative and service jobs, and that knowledge worker jobs might be the next step in that progression.
“One could argue that just as we automated manual labor jobs in the 18th and 19th century, administrative and service jobs in the 20th century, that the 21st century is where knowledge worker jobs really start to take it on the chin,” he said. “I think there is some sense of historical inevitability about this that we have to address seriously.”
Davenport identified several technologies that are driving knowledge work automation, including analytics and big data, machine learning, artificial intelligence/deep learning, and cognitive computing. He said that as analytics has evolved, it has become more recommendation oriented.
“Now I think it’s important to add a set of automated analytics at the top that says we are not just going to help you figure out answer, we are going to take action on it,” he said. “We are going to make a decision and we are going to forge ahead with the action related to that decision. There are all sorts of spheres in which that is already taking place.”
Davenport identified 10 knowledge work jobs that he called automatable: lawyers, accountants, radiologists, reporters, marketers, financial advisers, architects, teachers, financial asset managers, and pharmaceutical scientists.
Augmentation Instead of Automation
As another possible result of the ongoing evolution of technology, Davenport offered augmentation – humans and computers working together to make better decisions – as an alternative to automation, in which technology simply takes over the jobs of humans.
“Augmentation means humans are helping computers make better decisions, and vice versa,” he said. “People do this by aiding automated systems that are better at a particular task or by focusing on tasks at which humans are still better. It’s an ever-changing domain.”
Currently, this cooperation of humans and machines can produce results better than either computers or humans alone. Davenport offered the classic example of freestyle chess and a 2005 freestyle chess tournament in which two amateur players using three laptops defeated both grand masters and supercomputers.
Five Possibilities for Augmentation
Davenport offered five steps to augmentation in jobs:
Step in. Learn the system, how it works, its strengths and weaknesses, and how and when to modify it.
Step up. Monitor the big-picture results of computer-driven decisions, and decide whether to automate new decision domains.
Step aside. Focus on areas that people do better than computers, such as the creative and interpersonal.
Step narrow. Focus on areas that are too narrow to be worth automating.
Build the steps. Create the automated systems.
Davenport also offered advice for knowledge workers who are concerned about being displaced by automation and how to become an augmenter.
- Understand the ins and outs of how computers do your tasks, and try to improve them.
- Specialize in a component of your job that can’t be done well by a computer, such as sales.
- Write computer programs and algorithms yourself.
- Find a narrow job niche that no one would bother to automate.
Davenport ended by implying that understanding what it takes to become an augmenter – learning, changing what you do, and a lot of work to make it happen – may determine whether you keep your job.
“There’s always the artisanal plumber route,” he said. “All those knowledge workers could get some really artistic plumber’s helpers and go to work.”