Gigaom brings you our unique analysis and commentary on the present and future of AI.
In one of the classic Sherlock Holmes mysteries, "The Adventure of the Resident Patient," the story begins with Dr. Watson looking around and we're privy to his thoughts as he's sitting in the 221B Baker Street address with his glance going from one place to another. Then out of the blue Holmes answers a question that Watson had only ever thought.
At first, Watson is just flabbergasted by this and asks, was Sherlock Holmes able to read his mind? Sherlock Holmes couldn't read minds, of course, and he explained to Watson that the series of looks at a photo on the wall, out a window, and Watson's expressions allowed him to construct a picture of what it was that Watson was thinking, and then be able to enter into that mental conversation.
I'm a doubter about many things that AI is supposed to be able to do, but I do think that it will be able to do this. It will become so familiar with us--how we look things up and what we want to know and what we do next--that it can essentially read our minds, track where our eyes go and what we look at and what words we use and all of the rest. That won't require an AGI by any means. It will simply require an AI that's trained on us.