Ex Machina: AI and the Future of Supply Chain Planning
There’s been a lot of discussion lately about the future of artificial intelligence and its impact on human beings. It recently gained heightened attention with the movie “Ex Machina,” where an AI creation outsmarts its maker, escapes the closed experimental environment, and enters the real world (ex machina is Latin for “out of the machine”).
The movie popularizes concerns expressed by technology pundits like Elon Musk and Bill Gates, and philosopher Nick Bostrom. Bostrom argues that an “intelligence explosion” could enable AI to advance itself so exponentially that it exceeds the intellectual potential of the human brain by many orders of magnitude.
So as a leader in applying machine learning in supply chain planning, we thought it would be interesting to consider what the future of AI means to supply chain planning. Here goes.
In today’s complex demand and replenishment environment, the need to consume and leverage increasing amounts of data clearly favors the use of more machine intelligence—to better sense and shape demand, and to adapt supply to shifting consumption and replenishment needs. IBM says we now produce more than 2.5 quintillion bytes of data daily, 80% of it unstructured and “invisible to current technology.” IBM promotes “Cognitive Everything” organizations that use business analytics “to discover insights into their performance and identify future opportunities, … find correlations, create hypotheses, and remember, and learn from, the outcomes.”
Consulting firm McKinsey concurs. In an article entitled “An Executive’s Guide to Machine Learning,” McKinsey says, “The unmanageable volume and complexity of the big data that the world is now swimming in have increased the potential of machine learning—and the need for it.” McKinsey says that what machine learning “does extraordinarily well—and will get better at—is relentlessly chewing through any amount of data and every combination of variables.”
I see this broad trend applying to supply chain planning. I have addressed the clear trend towards the use of machine learning in numerous blog posts:
- In “Rise of the Machines—Predictive Analytics in Supply Chain Planning,” we wrote, “Machine learning, genetic algorithms, and other types of artificial intelligence are now in supply chain planning software packages, yielding powerful results.”
- In “Coming Now—The Age of Advanced Demand Analytics,” I described how one of our CPG company leaders has added analytics with machine learning capabilities to analyze their demand planning—gains included a rapid 20% reduction in forecast error and a 30% reduction in lost sales.
- In “Five Things You Need to Know about Machine Learning for Supply Chain Planning,” I quoted Gartner’s Noha Tohamy—“One of the defining characteristics of machine learning is uncovering new interdependencies previously unobvious to the user.”
Gartner also predicts supply chain planning is moving in this direction. In their recent “Predicts 2016: Reimagine SCP Capabilities to Survive,” the research firm says that in their recent survey supply chain organizations expected the level of machine automation in their supply chain processes to double in the next five years.
In the supply chain planning future I see, planners move from “in the loop” to “on the loop” managing the planning process at arm’s length. In a previous blog, I compared this to highly automated process control rooms where operators no longer need to make continuous adjustments, but instead let the automation handle much of the day-to-day effort. We have seen examples of this with our advanced customers such as Costa Express, which is now managing more than 5000 points of sale with just 2 planners.
James Cooke of Nucleus Research calls this the “Dark Cockpit” concept (“Dark Cockpit Lights up the Supply Chain“). “In the 1980s, due to the proliferation of instrument gauges on the airplane console, researchers streamlined the cockpit design to make it easier for the pilot to fly,” Cooke says. “When the plane is operating properly, no lights are illuminated on the instrument panel. When a light goes on, it alerts the pilot to a problem in the making.”
Similarly, he contends “the next generation of analytics in supply chain software under development now—predictive and prescriptive analytics—will necessitate the use of dark cockpit principles. These advanced forms of BI… make it possible for supply chain managers to gain insights in real time to their current operations, anticipate issues and take corrective action before a problem arises.”
Planners must still have the final say. “Humans are more capable of understanding the upside and risk associated with a decision, making a judgment based on quantitative analytical findings and qualitative business considerations,” Tohamy says. “Finding the right balance of human-based and machine-based approaches will offer the organization the speed and accuracy of science with the art of supply chain domain expertise.”
In the end, I see this technology evolving not unlike others that preceded it—into a combination of human and machine skills, leveraging what each does best, to increase productivity and free planners from the daily crunch to manage exceptions, think forward, and focus on value-added activities. Both companies and planners are much better off.