Artificial intelligence is everywhere. In our Google searches, in our mobile banking apps, in our thermostats and refrigerators and endless requests for Alexa. At work, AI tells sales reps which accounts they should be pursuing and helps lawyers instantly analyze piles of contracts.
Is it any wonder that we’re starting to think it might be OK if the machines take over?
A recent global survey found that 64 percent of more than 8,000 respondents said they didn’t just embrace AI — they would actually trust it more than their manager.
Tony Deigh, chief technology officer at the Cambridge machine-learning-based employment platform Jobcase, understands this impulse. As AI gets better at recognizing complicated patterns from huge troves of data, it could conceivably be applied to many roles, like being a boss.
“Would I take career advice from a machine? Sure,” Deigh said. “If it’s explainable.
“I don’t think we’ll get to human-like intelligence, but in a lot of dimensions . . . we can come pretty close,” he said.
People’s comfort levels with AI seem to be growing as they become more exposed to it. Half of the respondents in the survey, conducted for the software giant Oracle and the executive development firm Future Workplace, said they used some form of AI at work, up from a third the year before. Compared to the previous year’s survey, uncertainty about AI is down, and excitement is up.
Artificial intelligence — computer systems that can perform complex tasks that usually require the human brain, such as decision-making and speech-recognition — has become far more powerful in the last few years. Machine learning, a technique driving much of AI in which computers teach themselves to find patterns, is also advancing rapidly. One recent Harvard Business School study even found that AI was better than humans at predicting which jokes people would find funny.
Artificial intelligence is invading the workplace at all levels, from factory floors to high-rise office jobs. People with bachelor’s degrees are expected to be exposed to AI at more than five times the rate as workers with high school diplomas, according to new research from the Brookings Institution.
Tulip, in Somerville, offers a platform that allows manufacturers to build apps to assist workers, including computer-vision-powered tools that provide assistance as workers assemble parts, as well as detecting movement and speed to help supervisors determine who needs more training. Workers are often suspicious at first, said chief executive Natan Linder, asking: “Are you guys trying to track us? Are you going to replace us?” But when employees’ performance improves, he said, they have no interest in going back. “Would you give up your computer [if] your boss gives you a typewriter?”
Technology that isn’t technically AI but guides workers through their jobs is also easing workers’ fears about answering to a computer. At Dentsply Sirona in Waltham, which manufactures custom components for dental implants, senior process engineer Dan Ron uses Tulip technology to digitize the workflow, including bins that light up to show workers where to find the right part.
Does Ron trust machines more than he trusts humans in some cases? “Oh yeah,” he said, laughing. “I know that this is repeatable. It’s going to constantly give me the results I’m going to expect, whereas memory fails. Even the best people can have a bad day.”
Yinaira Valentin, a 19-year-old college student and aspiring lawyer, works at a DHL Supply Chain warehouse in Wilkes-Barre, Pa., packaging apparel and accessories for an outdoors company. She said she was immediately comfortable with the AI robots — made by Locus Robotics in Wilmington — that help her locate products because they made her job easier.
“I feel like I can trust them,” she said. “I pick the wrong item and I don’t even notice until the robot tells me.”
But would she want it to be her manager? Maybe, she said. Since the technology she works with is made by humans, she wouldn’t necessarily feel like she was being controlled by a machine: “I don’t know if technically it would be robots, but if it were, that would be cool.”
Workers still want human managers, however, said Emily He, a senior vice president at Oracle. They just want them to focus on what they do well, such as understanding feelings and providing leadership, and leave tasks such as maintaining work schedules and balancing budgets to the machines.
Yet all this trust in artificial intelligence might not be deserved. Julie Shah, a roboticist at MIT, cautions that the anthropomorphic nature of artificial intelligence — its ability to carry on a conversation of sorts, like Siri and Alexa — might be engendering trust it doesn’t deserve. The people who program these machines, and the data that drive them, are baking in human biases, she said.
“Just because people say they trust these systems does not mean that they should trust these systems,” she said. “The more aware you are of the technology, the less you trust it, because you know too much.”
At Jobcase, for instance, engineers noticed that women weren’t clicking on descriptions for UPS package handlers, but when they changed the description to warehouse worker, it garnered more responses from women. It wasn’t that women didn’t want to work for UPS, it was that the original ad was using a term that was unappealing to them. If the situation had been left up to a machine-learning tool, said Deigh, the company’s chief technology officer, it likely would have just stopped showing UPS jobs to women. And no one would have realized it.
“It’s actually not so much that my faith in the technology has gone down, but my concerns about it have gone up,” Deigh said. “My fear is not that I’m going to be bowing to a robot overlord soon, it’s that humans are, often with good intent, going to apply these very powerful tools without thinking through the potential consequences.”
And even those who fully embrace and trust AI don’t necessarily want it as their boss.
Boston’s Shawmut Design and Construction uses software to analyze photographs from 200 job sites to identify potentially hazardous conditions. The system, dubbed Vinnie and developed by the Cambridge startup Smartvid.io, is “bettering life,” said Shaun Carvalho, vice president of safety. But he’s not ready to let Vinnie call the shots.
“I trust the data that Vinnie gives me, but it’s data, that’s all it is. To me it’s similar to a calculator,” he said. “Would I trust a robot to come up to me and tell me how to lay out a wall? Not today. I’d rather have that be a person.”
Lawyers are also using AI to comb through contracts and years of e-mails related to their cases. It’s making lawyers faster, better, and more efficient — “bionic” even, said Anne Stemlar, who oversees legal technology for Goodwin Procter out of its Boston office.
But neither Stemlar or her colleague Michelle Briggs, who focuses on electronic discovery in the litigation department, can imagine relying on technology alone to run the show.
“I don’t see a world with robot lawyers,” Briggs said.