The New Business Case For AI

At noon, on Friday, May 25th, 2018, at the Viva Technology event in Paris, France, I will debut a new perspective on AI. Here’s the punchline: the approach generally applied to measure the business value of AI needs an update. Read on for a primer on the new business case for AI.

The world is fascinated by sci-fi threats of robot armageddon, but what’s unfolding with artificial intelligence is much less cinematic. Companies that want to grow in today’s always-on, digital economy must commit to pervasive technology innovation, continuously exploring new ways to improve, optimize and re-invent through wave after wave of emerging and advanced technologies. The newest wave of innovations to tackle are robotic, intelligent and autonomous systems.

When applying AI specifically, there are at least five, non-discreet value propositions. They are to 1) reveal deeper insights that support decision making, 2) leverage human-like cognitive capabilities, such as to see, learn, engage and navigate, to enhance user experiences, 3) deploy new methods that sustain trust, 4) optimize system performance through learning in supervised and unsupervised ways, and 5) harness the full spectrum of automation technologies to transform business operations. While much of the focus today is on use-cases, the how and where to apply AI, there is a more basic and critical question that is often too quickly overlooked - “why” invest at all? What is the real business imperative for AI? 

The traditional answer is a business case based on incremental gains in process efficiency and productivity, but this approach is rooted in decades of investing in enterprise IT where incremental gains in human performance are celebrated. Said differently, companies often measure the impact of enterprise technology on business performance based on how it incrementally improves workforce productivity, but this framework fails to capture the full transformative potential of AI. For example, if a company implements an IT system, it may increase the number of geological images a scientist can review in an hour; or it may increase the number and accuracy of blood samples a lab tech can process in a day; or it may improve the volume of transactions the fraud department can monitor in a week. These parameters of performance are expressed in terms of what a person or team can accomplish within a single earth cycle (hour, day, week, etc.).

The challenge is that such a performance model under-scopes the level of benefit that can be realized through AI. It is based on both real and perceived constraints that simply do not apply to machines. Of course, machines are subject to the laws of nature, including thermodynamics and gravity, so components do wear down over time. Software can also get corrupted and fail. Nevertheless, AI does not sleep, eat or take vacations, so it can operate, which in this case, involves the emulation of human-like cognitive skills, beyond the limits of biology and earth cycles. As such, we must re-imagine how we think about performance, value and the future of competition.

Illustratively, an AI can acquire skills at non-biological speeds, experiencing life-times of knowledge in days, which it can then apply towards a complex problem, learning, improving and ultimately achieving superiority in the execution of that skill that may be fairly characterized as alien-like. This exact scenario has already played out, literally, in the gaming world on multiple occasions. 

In a recent article in the Wall Street Journal (March 2, 2018), retired U.S. Marine Gen. John Allen was interviewed about the role of AI in US national defense. He observed the inevitability of “hyper-war,” a scenario in which “AI [speeds] up warfare to a point where unassisted humans can’t keep up.” Putting aside the military context, a form of “hyper-war” is in fact emerging across every segment of the global economy and in every conceivable workplace setting from HR departments, call centers and hospitals hallways to warehouses, theme parks and retail shops. As companies integrate human workforces with robotic, intelligent and autonomous systems, the result is more than increased efficiency and productivity, it is an operating model no longer defined by what people can do, but what people and machines can do together. The new business case is therefore an operating model that includes and is directed by humans, but is nonetheless non-biological in its potential. Machine-scale performance introduces a new outer boundary of what’s possible, and changes the very basis for competition.

As companies scale AI, humans still matter, but human-scale metrics become less relevant. The future of performance, value and competition will be increasingly defined by the outer boundaries of computational infrastructures that automate human-like cognition in tandem with (and dependent on) human managers, modelers, trainers, designers and curators. If a company and its management does not fully grasp this concept, then it is possible they will not properly plan for, accurately target and fully capture the maximum value of AI.