The near-term future of AI: intelligent thoughts or intelligent actions?
Sir Tim Berners-Lee spoke at the Cannes Lions festival and shared his views about the impact artificial intelligence (AI) will have on the world. While there’s little doubt that AI will affect various aspects of our lives moving forward, let’s take a quick look at how it’s impacting us today, and how the future might look with a larger serving of AI.
The interesting truth is that most of us are already using AI daily via everyday tools (and toys) including Siri, GoogleNow, and Amazon’s Alexa-powered Echo. While these applications are exciting and entertaining in their use of AI, they’re also limited. They hear and process our requests and commands, but they don’t attempt to understand or truly apply intelligence. If you go beyond the scope of pre-defined device control commands, they fallback on web searches to supply the answers. So if you ask GoogleNow to “get tickets for Jurassic World,” it will show you the screening times at local theaters via Flixter or the Cineplex website, but it won’t automatically book tickets at your favorite theater for the 9:00PM show. Your request is converted into actionable data, but it isn’t understood in the truest sense.
Today’s AI doesn’t require machines to think. In fact, we’re quite a long way from there. But then, do we really need our machines to think intelligently, or do we need them to simply act intelligently? That’s really what we’re working toward -- creating algorithms that allows for a complex set of outcomes and apply some sort of intelligence that leads predictable actions on the other end. That’s what we can expect from AI in the near-term.
Ultimately, we expect AI to be able to mimic intelligent behavior, but let’s be clear: we can’t expect AI to mimic the intelligence of a human. It’s not capable of true human response.
One of the reasons we’re not seeing intelligent behaviors yet is because today’s applications are too general in their goals. In trying to accomplish too much, they leave a tremendous gap between the request and the rendered solution. Responses are, out of necessity, based in one domain at a time. If we want to see more intelligent applications sooner, the smartest route would be to tie the application to a single domain.
Let’s think back to the previous example of movies. If an AI-driven app was tied exclusively to a site like IMDb, imagine what it could accomplish once it learned your preferences. It could potentially order your tickets to the movie of your choice, book your optimal screening time at your favorite cinema -- knowing that your preference is the one with the lounge seating and not the closest one. It would know whether you preferred 3D or 2D, and that you like buttered popcorn and soda bundled into your ticket. It might even block off your calendar and send you a reminder when it’s time to leave.
We’re headed in that direction -- soon – and advancements in Natural Language Processing (NLP) is what will help get us there. NLP is a subset of AI, and it’s the surface layer that allows a machine to “act” more intelligently. By focusing development in specific domains, NLP can go deeper in deciphering the nuances of the spoken and written language. It operates in concert with backend machine intelligence to engage in decision-making on your behalf once it learns your habits and preferences without having to repeat the requirements with every interaction. So it would behave as though it “knows you” and sometimes it can even make better decisions for you than you do, because it has the ability to analyze vast amounts of data in an instant.
If algorithmic decision making is going to drive the future of business as Berners-Lee predicts, it certainly can drive our personal decisions as well. The extent to which that happens will depend heavily on the data we choose to provide to fuel those algorithms. The more specific and personal, the better as far as I’m concerned. As Berners-Lee points out, data has more value to me than to anyone else, and powered by my personal data, an AI app can become a very efficient personal assistant.
Charles Lee is co-founder of Genee. He tweets @chasylee