From Automation to Autonomy: Minimizing the Productivity J-Curve in Artificial Intelligence Adoption

Abstract

The productivity paradox is not unique to a single technology revolution, with the take-up of artificial intelligence having a similar pattern to that of previous information technology revolutions. Despite improvements in task-level productivity in knowledge work, macro-level productivity statistics for the economy are not increasing. There is a disconnect between the siloed growth of technology and macroeconomic growth. Organizations are also prone to replicate the fragmentation, creating "islands of automation" rather than autonomous applications that can orchestrate the execution of knowledge tasks. The productivity J-curve assumes that investment in capability development, redesign of workflows and processes, and alignment of architectures will have a negative productivity payback if the investment in the productivity gains has not returned. The length and depth of the productivity J-curve can be reduced through systematic reuse of the lessons learned in the evolution of IT (end-to-end workflow management, development of complementary organizational capabilities, and convergence of the pilots into architectural platforms). Cloud-native integration and distributed orchestration patterns help achieve this autonomy by enabling data-powered solutions that orchestrate and distribute autonomous decisions across enterprise systems to create micro-level efficiencies that lead to system-level productivity.