“Artificial intelligence has the potential to reinvent the world, from how businesses operate to the types of jobs people hold to the way wars are fought. In health care, AI promises to help doctors diagnose and treat diseases as well as help people track their own wellness and monitor chronic conditions. Watson’s struggles suggest that revolution remains some way off.” ~ Daniela Hernandez and Ted Greenwald, “IBM Has a Watson Dilemma,” Wall Street Journal, 11 August 2018.
People tend to have one of three responses to the abundant and persistent coverage of emerging technologies in business:
- Enthusiasm and optimism,
- A paralyzing fear of having dominant robot overlords,
- An old-fashioned eyeroll.
What all three reactions share is a sense of inevitability, an acceptance that full automation and digitalization are on the horizon and that their impact will be broadly disruptive.
But AI can’t stage its own revolution. As the IBM Watson team has learned in their efforts to apply the famous question-answering computer system to the enormous challenge of curing cancer, roadblocks persist – and getting past them will require the collective intellect of programmers and users. This creates a huge opportunity for humans.
Imaginative business leaders dream of applying AI to spend data analytics, achieving full supply chain transparency, and deftly automating internal buyer support. While these remain on the table as potential applications for AI in procurement and supply chain, AI is being held back like many of the other enterprise technologies we implemented with equally lofty goals.
If we want to harness the exciting possibilities of AI, we need to learn from the lessons of the past. Although AI is a ‘smart’ technology, the complexities of implementing and applying technology remain the same. Humans must be in the driver’s seat because the messy realities of data and business objectives still defy AI’s limitations.
Data is non-homogeneous and highly fragmented
Computer programs only know what they are told or what they can measure directly through sensors. If we don’t provide them with complete, representative data sets that clearly align, they will return answers founded on an incomplete view of the world. Because enterprise applications such as ERP systems struggle to earn widespread regard as the ‘system of record’ or ‘single source of truth’, supply chain has a data problem to solve before we can benefit from the promise of improved insight and analytics.
There is no cure for unpredictability
One of the challenges faced by the IBM Watson team in the context of cancer care is the volume and rate of new information resulting from clinical research. The same challenge exists with extended global supply chains and the regulations and geopolitics they are subject to. When the goal of analysis is based on complex math, AI is ready to take on the job. Explaining the quantitative impact of shifting tariffs or the unclear application of broad regulations to even the most intellectual thinking system will prove to be much more difficult.
Improvements must be measurable
In healthcare as in business, decision makers want to know what the ROI will be from major technology investments. Watson has gotten “good feedback” from physicians and its recommendations agree with human decision makers “most of the time”. Neither of these seems to justify the extreme investments of time and money required to implement a highly complex system. We don’t have to cheer against humans in this debate to get an ROI, but AI will have to deliver significant opportunities that would otherwise have been missed – or become far more cost effective – before robots start running our supply chains.
Whether you are an enthusiast, a skeptic, or an eye-roller, the best response to the possibility of AI deployment is to continue learning about its applications. For now, the job of getting supply chains and other complex systems is one that humans must take on. Without us, AI will never be able to reach its full potential.
Header photo: Sergey/fotolia.com