Human workers report feeling most productive when led by artificial intelligence

BY Charles Pulliam-Moore  August 25, 2014 at 2:07 PM EST
Photo by Flickr user Scania Group

Photo by Flickr user Scania Group

Researchers from MIT’s Computer Science and Artificial Intelligence Lab found that teams of human workers were at their happiest and most productive when their tasks were directed by robotic artificial intelligence.

Recognizing the value proposition provided by automated workers, the team, led by CSAIL student Matthew Gombolay, approached their research with the goal of harnessing a machine’s efficiency while still making use of human labor.

A team consisting of two humans and one robot were configured into three organizational models in order to compare how best to allocate different duties. All tasks were assigned by humans in the manually structured group, while the fully autonomous group featured tasks doled out only by robots. A third, semi-autonomous group split the difference, allowing one human to assign their own tasks while the other human was instructed by artificial intelligence. Each team was responsible for the gathering and assembly of specialized parts meant to be put together in under 10 minutes, mimicking a typical manufacturing setting.

Gombolay’s team found that participants in the study reported feeling at their most efficient and effective when working as a part fully autonomous group. While the human-led teams were as capable at assembling as their robot-led counterparts, the algorithms used by the robot’s artificial intelligence proved to be more effective at dealing with unexpected obstacles that could potentially slow down production.

While the conversation surrounding the role robots will play as a part of the manufacturing workforce typically focuses on the ways in which human workers could be made obsolete. This research, Gombolay said, could lead to situations in which human employees could be empowered by machines, rather than replaced by them.

In the future, Gombolay says, the scheduling and coordinating algorithms used in the experiment can be put to broader applications outside of manufacturing, including construction and search and rescue missions.