Paul Blase, Global and US Data and Analytics Consulting Leader, PwC Advisory Services
Many of the recent successes in AI have been in domains that are closed-ended (e.g., games) or tasks that are narrow (e.g., object identification). While definitely useful they lack some of the key characteristics of business decision making where business executives have to make decisions with often uncertain, incomplete, and inconsistent data in open-ended situations (e.g., should we develop a new product or enter a new market).
We draw upon the recent advances in AI and our own expertise in systems thinking, multi-agent systems, machine learning, natural language processing and simulation modeling to illustrate how we can help business executives make more open-ended strategic and operational decisions. Building intelligent agents that capture the initial (potentially conflicting) mental models of human decision makers and embedding them within a more complex model of the business environment allows for realistic and granular simulation of ‘what-if’ scenarios that iteratively improves the mental models of both humans and the intelligent agents that encapsulate their decisions and actions. Hence, we call this type of AI as Augmented
Intelligence where both the human and the AI progressively improve their performance by teaching and learning from the other.
Over the past decade we have helped business executives make some critical decisions using AI. We discuss three case studies – a large auto manufacturer making strategic go-to-market decisions on ride sharing and autonomous vehicles, a large financial services firm considering a decision in 2012 if they wanted to enter the mobile wallet market or not, and an oil and gas service provider using sensors and internet-of-things to make operational decisions on predictive maintenance of the oil drilling equipment. All three case studies involve humans working with AI to make better strategic and operational decisions.
We conclude by summarizing the lessons learnt from these client case studies and outline an approach for deploying AI for strategic and operational decisions.