Passing Go: Augmented Intelligence and the Improvement, Not Replacement, of Us 

(PHOTOCREO Michal Bednarek/Shutterstock)

Recently I had an interesting exchange about the rise of artificial intelligence and its implications. The discussion was centered around Google's AI-fueled AlphaGo program, which trounced Lee Sedol, the world's best player of the strategic game of Go. Since AI can now dominate Go, can we extrapolate that algorithms will be able to make critical business decisions and that AI will dominate business someday?

This extrapolation is not only wrong, it’s dangerous. A more accurate and responsible view is that today's advanced cognitive computing tools will augment, not replace, human intelligence.

Augmented intelligence is dedicated to developing tools that help humans sift through massive databases of information to find insights that are presented visually to help find new clues and help apply creativity to solve hard problems. It is expressed in machine learning, cognitive computing, and visual analytics, with the goal being human enhancement as opposed to computational sentience.

Artificial intelligence, by contrast, puts the computer in charge. Wikipedia describes AI this way: “In computer science, an ideal ‘intelligent’ machine is a flexible rational agent that perceives its environment and takes actions that maximize its chance of success at some goal.”

Mark Palmer of Tibco

Two examples illustrate the superiority of augmented versus artificial intelligence. The first comes from the world of chess. In 1997, IBM’s AI-driven Deep Blue computer beat Gary Kasparov, the greatest chess player at the time. What many don't know is that after Kasparov lost, players were allowed to enter chess competitions in partnership with computers. The result? Even average players, when augmented by a program, beat the most powerful computers using AI alone.

Moreover, when chess tournaments opened participation to groups of humans working with computers, we learned that human collaboration, great computer interfaces, and the decision-making process was far superior to the best AI by itself.

Kasparov himself summarized these stunning findings: “Weak human + machine + better (decision making) process was superior to a strong computer alone and, more remarkably, superior to a strong human, a machine, and inferior (decision making) process.”

The point being: it is good interaction between humans and computers – computers operating in support of human decision making – that makes for the best results. That’s what augmented intelligence is all about.

Do these findings apply to serious business? Yes.

On Wall Street, computer-driven trading volume over the past decade has risen from less than 5 percent to more than 80 percent. Trading is one of the most algorithmically driven application areas in the world. It’s the ultimate Petri dish for algorithmic innovation and research — great computer-driven decisions are worth millions.

Yet heads of electronic trading on the Street all agree: On non-volatile days, algorithms win. On volatile days, humans win. Every time.

Why? Computers are incredible at evaluating statistically significant patterns based on history. And thanks to modern computational power and streaming analytics, they can detect those patterns on billions of data points, far exceeding the capacity of the human brain to crunch data. But the key qualifier is “statistically significant patterns based on history.” The presumption is that the past can always predict the future.

On the other hand, the real world is filled with irrational behavior and emotion. As every great trader knows, the impact of human emotion and irrational behavior presents the greatest opportunity. Thus, it is the combination of intuition, creativity, statistical number crunching, and pattern discovery that wins the day. In essence, algorithms thrive on analyzing consistency across events and making decisions on the historical data points (non-volatile situations), while human users readily apply intuitive thinking and rationale during unpredictable or irregular events (volatile situations). The trick is to optimize the interface between humans and algorithms to achieve dominance for any situation.

Human emotion by definition is irrational, and that is where computation falters. I am not alone in this interpretation. Stanford University’s just-released One Hundred Year Study on Artificial Intelligence (AI100) report states: “The field of AI is shifting toward building intelligent systems that can collaborate effectively with people, including creative ways to develop interactive and scalable ways for people to teach robots.”

What the world needs now is love, sweet love, between humans and computers. Smart business leaders evangelize harmony between humans and analytics to augment team intelligence, rather than replace it. That is a worthy extrapolation.

Mark Palmer is senior vice president and general manager of analytics engineering at Tibco Software.