(Georgia Institute of Technology, 2019-07-29)
Clarke, John-Paul B.
Machines were initially designed to take over manual tasks. Over time, as sensors and algorithms became more capable, decision-making has been automated via prescribed rules. Further, when these rules can be verified a priori, automated systems are allowed to operate autonomously, i.e., without human supervision. Some envision that in due time machines will be able to both make decisions and operate autonomously. That said, despite science-fiction-inspired anxieties, it is unlikely that machines will replace humans in their entirety. Rather, the future will be one where humans and machines work together in partnership to achieve performance that is greater than the performance they could achieve individually. Specifically, the optimal allocation of functions to humans and machines will be dependent on their relative strengths with respect to autonomous decision-making and autonomous operation. In this presentation, I will present a framework for determining the optimum allocation of functions to humans and machines; and provide specific instances where each of the four possible allocations is optimal.