Lessons Learned From IBM's Initial AI Overreach

As with every other overhyped tech innovation, users of artificial intelligence, especially big tech companies, are often those most likely to be blinded by the potential versus the reality.

IBM's various Watson initiatives are a case in point. The company thought it could grab first mover advantage, and with the recognizable Watson brand name. But with big company ambitions come audacious and often unachievable goals. After numerous misfires, IBM has now gotten over itself, is partnering with others while exploring the far less expensive but more easily attainable wonders of incrementalism. JL


Steve Lohr reports in the New York Times:

The best course is often to use basic A.I. tools to automate mundane office tasks in operations like accounting, billing, payments and customer service — and to avoid “moon shots.” One of (the) prime examples of a misguided moon shot was IBM’s initial efforts to apply its Watson technology to diagnosing cancer. That was a high-profile science project that proved more difficult, time-consuming and costly than anticipated. In a nod to reality, IBM announced its A.I. technology will run on the popular clouds of Amazon, Microsoft or Google as well as IBM’s cloud.

A step-by-step approach is the counsel for business in a new book by Thomas Davenport, “The AI Advantage: How to Put the Artificial Intelligence Revolution to Work” (MIT Press). Mr. Davenport, a professor of information technology and management at Babson College, is a seasoned expert in using digital data to streamline operations and spot opportunities. And he’s been at it since well before the current A.I. wave, which is both a technological evolution and a branding craze.

His advice to mainstream companies is that the best course is often to use basic A.I. tools to automate mundane office tasks in operations like accounting, billing, payments and customer service — and to avoid “moon shots.”

One of Mr. Davenport’s prime examples of a misguided moon shot was IBM’s initial efforts to apply its Watson technology to diagnosing cancer. That was a high-profile science project that proved more difficult, time-consuming and costly than anticipated, though IBM continues that work with several leading cancer centers.

Speaking of IBM, the company was out in San Francisco this week for its annual Think conference, which attracted more than 25,000 attendees. Its executives presented its Watson and cloud technology as the trusted path for companies across the industrial spectrum that want to embrace A.I.

IBM cannot compete head-on with Amazon, Microsoft and Google in the big-spending game of building out massive data centers to provide the infrastructure layer of cloud computing to one and all. So it is seeking to shift the competition.

IBM talks about “chapter two” in the cloud market. Increasingly, it says, companies will run not only new applications but also their legacy software on the cloud, either private clouds in their own data centers or on IBM’s cloud. The next stage of cloud computing, Virginia Rometty, IBM’s chief executive, told Jon Fortt of CNBC, “is going to be driven by the modernization of mission-critical apps. That’s our sweet spot.”

And in a nod to reality, IBM announced a Watson Anywhere initiative: Its A.I. technology will run on the popular clouds of Amazon, Microsoft or Google as well as IBM’s cloud.