There was no shortage of news about machine learning and artificial intelligence in 2016. But while some of the...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
stories showed how far the technology has come, others showed how far it still has to go.
Google's AlphaGo game-playing AI algorithm checked in on the positive side. It was able to beat the world's best players of the game Go, which is often compared to chess, but is in fact exponentially more complicated due to the large number of possible moves and the strategies that develop from managing so many potential moves.
Then, there was Microsoft's Tay Twitter chatbot. The idea was to showcase advancements in AI and natural language processing capabilities, but it very quickly learned from other Twitter users to disparage women and minorities. Microsoft was forced to decommission Tay, making it one of the year's biggest machine learning and AI busts.
Other headline events produced more mixed results. IBM Watson scored a top hit for its musical collaboration with producer Alex da Kid. The song, which was co-written by Watson, was the most downloaded song on iTunes for a week in October.
Many people also got excited about Amazon Go, a new grocery store the retail giant will operate that uses machine learning and sensors to identify items customers take off shelves and automatically charge their accounts, allowing them to skip checkout lines entirely. These stories generated excitement, but their long-term importance remains in question.
Analyst Adrian Bowles of Boston-based Storm Insights Inc. said machine learning and AI projects like these, as well as more common applications like Apple's Siri smart assistant, are helping the technology become more of an expectation in people's everyday lives. They expect their phones, cars and appliances to be smart. This has raised the profile of AI and machine learning considerably.
"I think of 2016 as the year of modern AI," he said. "We're now seeing AI for practical purposes, and we'll get to the point where people expect a certain level of intelligence in their products."
For Forrester Research analyst Mike Gualtieri, advancements in natural language processing and chatbots were among the brightest points for machine learning and AI in 2016, notwithstanding the flameout of Tay. He said, in general, the ability for algorithms to respond to natural speech in ways that seem conversational came a long way this year, to the point where these tools are now much more usable. This is partly why Amazon Echo and Google Home, two smart home assistants that users interact with through voice control, became relatively popular.
"There's been some cool stuff in deep learning, mostly around speech, so what's important is the progress we've seen," Gualtieri said.
That being said, he added that he's worried about these technologies getting overhyped heading into 2017. He pointed out speech capabilities are variants on machine learning, something you might call pragmatic AI. But when most people think about AI, they're thinking of what Gualtieri called pure AI. This is what we're all familiar with from science fiction movies featuring intelligent robots that can think and interact as humans do.
Confusing the two kinds of AI could lead to disappointment if people hear we now have artificial intelligence, expect truly intelligent machines and then see only applications with limited utility.
"There's confusion about what it really is," Gualtieri said. "This can lead to inflated expectations. If companies think of this as more pure AI, they're going to be let down."
Self-driving cars will be a test of public trust in machine learning algorithms
AI algorithms ease chatbot development
Data preparation is a big hurdle in developing machine learning tools