IBM insists that its revised AI strategy – paid-down, less world-changing ambitions – is working. Following the recent reform of IBM’s cloud and AI businesses, computer scientist Arvind Krishna, who became chief executive last year, was tasked with reviving the development.

But the glorious visions of the past were gone. Today, instead of being a shorthand for technological prowess, Watson is a suicidal example of the technical hype and hubris difficulties around AI.

The march of artificial intelligence through the mainstream economy, it turns out, will be more step-by-step evolution than catastrophic revolution.

Throughout its 110-year history and time, IBM has introduced new technology and sold it to corporations. The company dominated the market for mainframe computers so much that it was the target of a federal antitrust case. PC sales really came to a halt after IBM entered the market in 1981, supporting small machines as essential tools in corporate sales fees. In the 1990s, IBM helped adapt its traditional corporate clients to the Internet.

IBM officials saw AI as the next wave for the ride.

Mr. Ferrucci introduced Watson’s idea to his boss in 2006 at IBM’s research labs. He thought that building a computer to deal with a question-and-answer game was known in the AI ​​field as a natural language process, in which scientists could advance. Program to identify and analyze computer words. The second research goal was to advance techniques for answering automated questions.

After initial suspicion, Mr. Ferrero-Waldner assembled a team of scientists – eventually more than two dozen – who worked out of the company’s lab in Yorktown Heights, NY, about 20 miles north of IBM’s headquarters in Armonk.

The Watson they built was a room-sized supercomputer with thousands of processors running millions of codes. Its storage disk was filled with digitized reference works, Wikipedia entries, and electronic books. Computing intelligence is a brute force affair, and the Hulking machine requires 85,000 watts of power. The human brain, by contrast, runs the equivalent of 20 watts.