In 2017, AI won't just be for the nerdy companies. Machine learning can help with mortgage applications and bridge safety, too.
Get ready for AI to show up where you'd least expect it.
In 2016, tech companies like Google, Facebook, Apple and Microsoft launched dozens of products and services powered by artificial intelligence . Next year will be all about the rest of the business world embracing AI.
Artificial intelligence is a 60-year-old term, and its promise has long seemed like it was forever over the horizon. But new hardware, software, services and expertise means it's finally real -- even though companies will still need plenty of human brain power to get it working.
The most sophisticated incarnation of AI today is an approach called deep learning that's based on neural network technology inspired by the human brain. Conventional computer programs follow a prewritten sequence of instructions, but there's no way programmers can use that approach for something as complex and subtle as describing a photo to a blind person. Neural networks, in contrast, figure out their own rules after being trained on vast quantities of real-world data like photos, videos, handwriting or speech.
AI was one of the hottest trends in tech this year, and it's only poised to get bigger. You've already brushed up against AI: It screens out spam, organizes your digital photos and transcribes your spoken text messages. In 2017, it will spread beyond digital doodads to mainstream businesses.
"It'll be the year of the solution as opposed to the year of the experiment," said IBM Chief Innovation Officer Bernie Meyerson.
It's enough of a thing that some are concerned about the social changes it could unleash. President Barack Obama even raised the issue of whether AI might push us to adopt a universal basic income so people other than CEOs and AI programmers benefit from the change.
New AI adopters next year will include banks, retailers and pharmaceutical companies, predicted Andrew Moore, dean of Carnegie Mellon University's School of Computer Science.
For example, an engineering firm might want to use AI to predict bridge failures based on the sounds from cars traveling across it. Previously, the firm would have needed to hire a machine-learning expert, but now a structural engineer could download AI software, train it with existing acoustic data, and get a new diagnostic tool, Moore said.
AI should reach medicine next year, too, said Monte Zweben, chief executive of database company Splice Machine and former deputy AI chief at NASA's Ames Research Center.
That could mean fatigue-free bots that scan medical records to spot dangerous infections early or customize cancer treatments for a patient's genes -- tasks that assist human staff but don't replace those people. "Precision medicine is becoming a reality," Zweben said, referring to treatments customized for an individual to an extent that's simply not feasible today.
A similar digital boost awaits white-collar workers, predicted Eric Druker, a leader of the analytics practice at consulting firm Booz Allen. Assessing whether borrowers are worthy of a mortgage is a standardized process, "but humans are making decisions at every step," he said. In 2017, AI will be able to speed many of those decisions by doing some of the grunt work, he said.
Cars increasingly are becoming rolling computers , so of course the auto industry -- under competitive pressure from Silicon Valley -- is embracing AI. Companies like Tesla Motors offer increasingly sophisticated self-driving technology, but drivers still must keep their hands on the wheel. Next year, though, the technology will graduate out of the research phase, predicted Dennis Mortensen, chief executive of AI scheduling bot company X.ai.
"One of the dozen or so serious self-driving initiatives will roll out a truly fully autonomous feature," though confined to highway driving, he said.
Why is it getting easier? Google and Facebook in 2016 released their core AI programs as open-source software anyone can use. Amazon Web Services, the dominant way companies tap into computing power as needed, added an artificial intelligence service. The computers are ready with a few mouse clicks and a credit card.
But to Chris Curran, chief technologist of consulting firm PricewaterhouseCoopers, AI will remain confined to narrow tasks like recognizing speech. A general artificial intelligence -- something more like our own brains -- remains distant.
"Data science bots -- something you could ask any question and it'll figure it out -- are farther away," he said. It's the direction Google is heading with Google Assistant -- which arrived in 2016 in its Google Allo chat app, Google Home smart speaker and Google Pixel phone -- but it's far from the ultimate digital brain.
Tech companies will push the state of the art further next year. Among the examples:
And maybe we'll stop feeling like such dorks when talking to our phones and TVs. "The tech arbiters of style," Tepper said, "are pushing hard to make it easier for people to talk to their devices and look cool while doing it."