“For decades, technologists have teased us with this dream that you’re going to be able to talk to technology and it will do things for us,” said Apple’s Phil Schiller before introducing the iPhone’s smart assistant, Siri, at a press event in 2011. “But it never comes true.”
Tim Tuttle was already very familiar with this letdown. He had dedicated the first part of his career to artificial intelligence (AI)–earning a Ph.D. from MIT’s AI Lab and performing research on high-performance computing platforms–but felt disappointed with the discipline’s practical applications. “I expected this to become a more important part of the systems we build,” he says, “The technology just wasn’t good enough, the data sets weren’t large enough, and as a result software systems that used AI weren’t very smart at all.”
To Tuttle, Siri was just another example of the broken AI promise Schiller was describing. Though the feature looked impressive on stage, it required users to push a button in order to get its attention and waited until they had completed their questions to show whether it understood them. He still believed AI could do better. His startup, Expect Labs, was working on a different kind of smart assistant–one that acted much more like a human assistant.
Instead of waiting to be asked a question, the technology, now backed by Google, Intel, and Samsung, among others, constantly listens to conversations and serves up information before you ask for it. If it hears you talking about a trip you’re taking to Cancun, it might surface recent news articles about the city or a travel website’s review of hotels. If you’re telling a friend about a YouTube clip, it might go find that video for them. “These devices have microphones and cameras,” Tuttle says about our iPhones, smart watches, and Google Glasses. “They know where we are, they know who we’re with, they know the same things that another human being would know if they were in the room with us.”
That vision might seem creepy to privacy advocates, but Expect Labs has focused on its opportunity. The company just releasde its first app, a simple voice conversation tool called Mindmeld, but it also has a much broader implementation in mind. It plans to partner with developers who want to use the technology in their own devices or applications–whether those are smarter cars, online education programs that scale to answer everyone’s questions, or conference software products that automatically surface references materials. “This technology will be part of the plumbing of every single application you use,” Tuttle says. “We’re actually getting to the point where these computer programs can be kind of smart.”