By Dusty Weis, AEM Strategic Communications Manager

John Henry had his battle with the steam drill. Ken Jennings had his Jeopardy game show duel with IBM’s Watson.

Both men will live on in legend as the last of their kind to fall under the steady march of machine progress. Jennings, though defeated, at least survived his contest, and quipped, “I, for one, welcome our new computer overlords.”

Regardless, IBM’s Watson artificial intelligence (A.I.) platform is neither capable of nor interested in world conquest, according to Toby Capello, vice president of expert & delivery services at IBM Watson A.I.

“It wasn’t about playing a game or winning at Jeopardy,” Capello says. “In a lot of ways, it was the ultimate test for A.I. in that time in history. Over seven years, we taught a computer how to understand natural language, make sense of a question and bring back an answer.”

Capello explains that, in the years since Watson’s 2011 victory over Jennings, A.I. platforms have increasingly been using that skillset to solve business problems and create new efficiencies for manufacturers. And, in a presentation at AEM’s Annual Conference, Capello explained to members what that will mean for equipment manufacturers and the imperative to put that technology to work for themselves.

“A.I. is going to disrupt your industries, your markets, your customers and your competitors,” Capello says. “It’s not a matter of if, it’s when—and I truly mean that.”

 

What Is Artificial Intelligence?

“In its purest form, A.I. is a computer mimicking human behavior,” Capello says. “In most cases today, it’s mimicking cognitive behavior. It understands, reasons, learns and interacts.”

But how is that useful in a business case? Capello says that these directives enable users to turn their unstructured data into structured data.

A simplified example with which many readers might be familiar is the “Recognize Text” feature on Adobe PDF. A PDF, after all, is just an image of a document—light spots and dark spots.

But when you click “Recognize Text,” an algorithm looks at the image, understands that it sees letters on the page and reasons out what the text says. It was trained to do this by interacting with countless documents and learning from its mistakes, which were pointed out by programmers who “trained” it.

Of course, “Recognize Text” is able to process pages of documents in a few short seconds—orders of magnitude faster than the best human speed readers. And this, Capello says, is the real power of A.I.—the ability to rapidly organize unstructured data, find patterns and solve problems that would take humans decades to unravel.

How Does Artificial Intelligence Work?

If A.I. is a computing power engine, data is its fuel—and the more fuel you pump into that engine, the more results you can drive. Capello says most organizations are packed to bursting with decades of old data, and even a dusty, disorganized file cabinet that hasn’t been touched in 50 years can power results.

“Don’t wait for ‘good’ data; do something with the data you have,” Capello says.

Once you’ve identified an organizational problem and the data to help solve it, Capello says the next step is to “train” the A.I. to find solutions. Over the course of several months, the A.I. analyzes the data and attempts to reason out viable ones.

A.I. is guided in this process by a team of subject matter experts from the organization itself, who tell the A.I. when it has reached a correct or incorrect conclusion. The A.I. catalogues these successes and failures, learns from them and adapts its approach accordingly.

Using this method, Capello says companies have used the Watson platform to tackle problems ranging from complex maintenance solutions to automated customer service. But he cautions that it’s important to clearly define the scope of a problem before attempting an A.I. solution.

“The A.I. journey is littered with failures from companies that have tried to bite off too much,” Capello says. “They’ve gone after what might be possible, not what’s do-able. It was either something that didn’t clearly drive value, or they didn’t have enough data for it.”

How Can Equipment Manufacturers Benefit from A.I.?

Capello relates the potential benefits of incorporating A.I. business solutions with a number of IBM Watson success stories.

In one case, an Australian energy company used A.I. to tackle an unlikely design problem—birds roosting on the steel superstructure of its oil platform helipads and posing a flight risk. Armed with 28 years of operational and design data, Watson zeroed in on a decades-old report from a now-retired engineer who had discovered a support configuration that discouraged birds from roosting—an insight that could have been lost to the ages.

“The only way they could have done that otherwise is if that engineer was still around to ask, or if they had happened to look in the right file,” Capello says. “But Watson was able to pull it up right away.”

Capello says A.I. has applied similar approaches to answering customers’ questions in a call center scenario. One large equipment manufacturer sicced A.I. on the RFP response process, routing inquiries to appropriate staff and using past data to project pricing impacts. And, in the legal world, A.I. is being used to analyze class action lawsuits and draft response letters in a matter of minutes.

Ultimately, Capello says A.I. success is limited only by a user’s ability to define challenges and provide access to the data needed to solve them.

“A.I. is already everywhere today,” Capello says. “But this isn’t just about machines—this is humans and machines working together to solve problems.”

Subscribe to the AEM Industry Advisor.