Why Artificial Intelligence Is Going to Look Like a Smartphone Keyboard
(Thinkstock)
"The future of artificial intelligence is autocorrect," Ben Medlock, the co-founder at SwiftKey was saying. We were at a dinner during South by Southwest, hosted by SwiftKey and Evernote, and the discussion had turned to the standard nerd debate: When will we get real, usable artificial intelligence? And what will it look like?
It’s no surprise that Medlock said what he did. SwiftKey makes replacement keyboard software for Android and iOS devices. He was just dreaming big and extrapolating: SwiftKey keyboards can predict what you’re going type and, if you pay attention to what the keyboard is suggesting, you can type a lot faster. This is a standard feature in software keyboards, of course, and different software companies fight over the efficacy, personalization features, and intelligence of their algorithms.
Thinking about autocorrect and word suggestions as AI might seem reductionist, but another AI expert at the dinner, Stephen Wolfram, of Wolfram Research, has been saying this for a while now. We can make AI a “neural prothesis,” he says. It can be something that makes our minds stronger, that eases our mental burdens.
Wolfram expects that AI will indeed become like “auto-suggest” for your life. As he says, if you can let a keyboard on your phone suggest the words you probably want to type, why not also let your phone suggest what you might want to do? The software can base its suggestions on the vast and growing knowledge about you that it’s now very easy to gather.
Some of this is already here. Google Now will suggest that you wrap up your work and get ready to leave so you can get to your next appointment in time, and it gives you the alert enough in advance that you can actually get out the door when you should.
As we know from Apple’s latest announcement about ResearchKit, your phone can also monitor your health and physical state. Why couldn’t it suggest that you, say, take a nap (or get a cup of coffee) when your mental acuity is failing? Why not have an app suggest a menu item when you’re at a restaurant — one that balances health with your enjoyment?
Artificial Intelligence could be designed as self-serving and goal-seeking by itself. We could (and probably will) have AIs that are used for individual and social control, or to fight wars. But not all AI has to go this way. Evernote CEO Phil Libin prefers to call AI, “augmented intelligence,” and his company, he says, is using AI to help people do what they want to do: Live better lives, and be more intelligent. “In five years we will be smarter,” he says. Not because our brains will be functioning better, but rather because our software will be amplifying our minds, reinforcing our goals, and helping us with analytical thinking and the archiving of ideas.
"AI is prediction," SwiftKey’s Medlock says. And it’s prediction about us. More importantly, it’s prediction for us.
Read More: Drinking Coffee with Stop the Robots, SXSW’s Anti-Robot Protesters
(Disclosure: I worked at Evernote in 2013.)
