Richard Borge for the Boston Globe
Technology moves blindingly fast — except when it doesn’t. Take artificial intelligence.
This spring, IBM opened a new center in Cambridge for its “cognitive intelligence” software, also known as Watson. Five years after winning $1 million on the game show “Jeopardy,” Watson is ready to analyze vast quantities of data and spit out deep insights. All that intelligence, according to an IBM release from September, represents nothing less than “a new era in computing.”
Thirty years earlier, IBM was setting up another office a few blocks away to focus on the then-trendy realm of artificial intelligence. In the summer of 1986, IBM executive Herbert Schorr told an AI conference, “We look at artificial intelligence as an extremely timely technology that is ready for commercial applications.”
And 30 years before that, in the summer of 1956, academics and researchers from Harvard, IBM, and Bell Labs gathered on the campus of Dartmouth College to spend the summer making “a significant advance” on building computers that could use language and begin to solve problems as humans do. The term artificial intelligence had just been coined.
So when you read predictions about sentient robots taking care of Grandma, driverless cars causing unemployment to soar, and emotive software like Samantha from the sci-fi movie “Her,” some of it might remain in the realm of speculation for a while yet.
The most pragmatic thing you can say about artificial intelligence is, in the words of Howard Cannon, that it is a label slapped on “the set of problems we don’t currently know how to solve with a computer.” Some might get solved in years; others decades or more.
Cannon was a cofounder of a Cambridge firm called Symbolics, spun out of MIT’s AI Laboratory in 1980, that designed computers especially for artificial intelligence software. “A CEO would be reading Fortune magazine in the bathroom and it would say, ‘All the big companies are starting artificial intelligence labs,’ and he’d think, ‘I wonder if we have one of those.’ It was a little bit of a self-feeding phenomenon,” Cannon says.
But the artificial intelligence boom of the 1980s eventually went bust, and onetime AI high-fliers like Symbolics, Lisp Machines, Palladian Software, and Applied Expert Systems vanished. A Globe headline from 1988 proclaimed, “Hard Times in Boston’s ‘AI Alley.’ ” Not long after that, most people forgot about AI Alley entirely (it was a term used to describe the cluster of companies in Kendall Square on the edge of MIT’s campus).
These days, not only is IBM back in Cambridge, but the enthusiasm and funding for AI companies in Boston and across the country have returned — accompanied by a raft of buzzphrases suggesting that computers are getting brainier, from “deep learning” to “synaptic intelligence.”
But unlike the 1980s, when technology fresh from the academic lab stumbled out into the world looking for problems to solve, many of today’s companies are focused on specific business problems. Nara Logics of Cambridge, for instance, helps large companies anticipate whether “an event happening within your supply chain is going to become a problem — and what you can do about it,” chief executive Jana Eggers explains. The company says its software can also analyze financial transactions in real-time to spot instances of fraud that don’t necessarily fit historical patterns.
Talla, another Cambridge startup, aims to store some of the conversations happening inside of companies on messaging software like Slack and HipChat, and begin to build a map of who knows what in the organization. “If you’re a new employee trying to solve a problem or figure out who has had experience with a certain domain, Talla will try to guess who in the organization knows the answer,” says Rob May, the company’s chief executive. “In most companies, there’s no system of record of what is in people’s heads. That is what Talla is becoming.”
A Boston startup called Cortex is working with companies like Heineken and Ritz-Carlton to analyze the materials they publish on social media, and apply artificial intelligence to “predict how people will react and respond,” cofounder Matt Peters says. Looking at all of the material a company has shared in the past on a service like Facebook or Instagram, as well as what its competitors have posted, “We can say, ‘At Friday night at 7:30, you want to put these things into a picture, or create a post with these keywords,’ ” Peters says.
“AI can be a little program running in the background,” he says. “It doesn’t equal a Terminator robot or Data from Star Trek. Cortex knows how to do some things well, but it isn’t going to take over the world. It wouldn’t know how.”
Adam Honig, founder and chief executive of Boston-based Spiro, says he was inspired by “Her,” the 2013 film by Spike Jonze, about a man who falls in love with a Siri-like intelligence. Honig felt like salespeople might benefit from a personality that prodded them to follow up with prospects, based on data about the salesperson’s past performance. Users pay $12 a month for the software, and they can choose what personality they like best: an Andrew Dice Clay-inspired trash-talker, a rah-rah motivational coach, or a Kardashian-like gossip girl. Spiro has a part-time (human) comedy writer under contract to produce material for the system. “Humor is critical to our success,” Honig says, and AI apparently can’t craft jokes just yet.
Also worth tracking are GiantOtter and Semantic Machines, two startups trying to develop software that can hold up its side of a conversation with a human. One of the 17 PhDs employed at Semantic Machines is the former chief scientist for Apple’s Siri feature, Larry Gillick.
Cannon, the former Symbolics executive, says that while “there are always business opportunities to automate things that weren’t automatable before,” he still feels like truly sophisticated artificial intelligence still has “a ways to go.”
But bullishness in this corner of the tech world is recursive. “Why is it spring now after all the AI winters?” asks Vivjan Myrto, a managing director of the Boston investment firm Hyperplane, which has put money into Spiro and Talla. “AI techniques have come of age, and you’re seeing big advances in hardware. It’s a really historic moment.”