By Brian O'Keefe, assistant managing editor
FORTUNE -- In your new book How to Create a Mind, you say you've unlocked the secret of human thought. How long before we can build a functioning artificial brain?
On that question I've consistently said 2029. Both hardware and software are progressing exponentially. If logical thinking were the essence of intelligence, then computers are already superior to us. The areas where humans still have an edge are in our emotional intelligence. Emotion is not some sideshow or distraction to intelligence. Being funny, being sexy, expressing love in a convincing way -- those are the cutting edges of human intelligence.
In the book you introduce your "pattern recognition theory of mind." What does it tell us about how the brain works?
The main focus is on the neocortex, which evolved with mammals. It now comprises 80% of the brain in humans. One of the key themes of the book is that neocortex is neocortex. It all runs the same algorithm. This runs counter to the spirit of a lot of neuroscience, which is based on the idea that different regions of the brain run different algorithms for different functions. The neocortex is actually a module of roughly 100 neurons, and it's repeated over and over again. We have a total of about 300 million modules, and they are pattern recognizers. They learn to recognize a pattern, and they're organized in a hierarchy. That hierarchy is built from our own thoughts. So it's true that you are what you eat, but it's even more true that you are what you think.
When will the average person have his or her brain connected to and supplemented by computers?
One answer is that we have brain extenders already, even though the vast majority are not physically connected. Humans have always built tools to compensate for our limitations. We now have mental tools to extend our reach. So I can access all of human knowledge in sense with a few keystrokes and so can the kid in Africa with a smartphone. So these are very literally brain extenders.
To answer your actual question, I would say the 2030s. We will have intelligent computerized devices the size of blood cells, and we'll put them in our blood stream. They'll go inside our brain and provide gateways to the cloud directly from our brain. And one of the things we'll do will be to extend the neocortex so that we're not limited to just 300 million pattern recognizers. Consider the last time we added more neocortex, which is when homo sapiens evolved. That was the enabling factor that permitted the evolution of language and art and science and music and literature. It really came from this additional quantity, which made this qualitative leap possible. So 300 million is a lot in terms of the fact that it made this qualitative leap possible, but it's also very limiting, and we struggle with that limitation every day. Just confront how long it takes to read a book, let alone how long it takes to learn a new language.
Suppose you want a billion or 10 billion pattern recognizers for a few seconds because you're doing some kind of complex search. What happens now is it goes out to the cloud. The interesting things that happen on your cellphone or notebook are not happening in that rectangle, they're taking place in the cloud. And the cloud itself is growing exponentially.
Will increased processing power change the way we think, or just increase our memory and the amount of information we can access?
As adults, we've filled up our neocortexes. We're constantly needing to forget things to free up neocortex to learn new things of any magnitude just because of that physical limitation. Just imagine what the next expansion of neocortex capacity will permit qualitatively. But there are other advantages of a non-biological neocortex. It's not just additional quantity. Computations can be done faster. Also, information stored biologically is not backed up. We routinely back up everything on our non-biological devices. And you can download new knowledge to it. So we will become a hybrid of biological and non-biological intelligence. We're not going to be plugged in a la the Matrix. It's going to be WiFi. And it's not going to be one thing. You're not going to get a form: Check here, I want to be enhanced, Yes or No. There's going to be a million choices. There's a million choices now just of iPad apps. Go out to the late 2030s and there'll be some nano-bot software combinations that are very conservative that enhance your immune system and your memory that everyone uses. And there'll be some more cutting edge things that only certain people use. How many people today opt out of this whole connectedness? Maybe the Amish. So people generally don't opt out, and there's going to be different choices.
Is technology changing our brains now in an evolutionary way?
Not biologically. What does happen though is that our neocortex isn't going to bother using up a lot of its capacity learning things that it knows it can get through its brain extenders. There was a controversy when I went to college about these little devices called calculators. People said, 'Oh, kids aren't going to learn arithmetic.' That turned out to be true. They could probably still add and subtract, but doing long division became a lost skill. But the calculators didn't go away and Google (GOOG) isn't going away. And we don't need to fill up our precious neocortical real estate with lots of facts. The idea of rote learning is obsolete because we carry access to all of these facts and knowledge around with us. We do need to teach people how to use knowledge, how to solve problems and so on. But the brain extenders are not going way. People say, "These things are making us stupider." If you define intelligence as the ability to remember random facts that might be true. But in fact we are much more intelligent with our brain extenders. These powerful tools are making us smarter, but you have to include them as part of who we are. And in my view they are part of who we are.
A shorter version of this interview appeared in the January 14, 2013 issue of Fortune.
More from The Future Issue
Getting supercomputers like IBM's Watson to understand slang may be the final frontier in machine intelligence.
FORTUNE -- The scientific test to gauge if a computer can "think" is surprisingly simple: Can it engage in small talk? The so-called Turing test says a computer capable of carrying on a natural conversation without giving itself away can be considered intelligent. So far, no machine has made the cut.
Eric Brown, a research scientist MOREMichal Lev-Ram, writer - Jan 7, 2013 5:00 AM ET
Would that Apple's intelligent assistant were as reliable as its Christmas TV ad suggests
Apple (AAPL) has been pushing Siri hard as the key selling point for the iPhone 4S, especially on TV with ads like the Santa spot it began airing this week. (It's also available on YouTube.)
But our experience is that between system outages, misunderstandings and the basic limitations of its knowledge base, Siri is considerably less reliable than MOREPhilip Elmer-DeWitt - Dec 20, 2011 7:21 AM ET
The downside of server-based voice activated computing
Apple's (AAPL) Siri, to paraphrase Alan Kay's comment about the original Macintosh, is the first voice-activated artificial intelligent assistant good enough to criticize.
Good enough, in fact, that Asymco's Horace Dediu has suggested that voice-activation might be the next revolutionary user interface, as disruptive for future computing devices as the mouse, the scroll wheel and the touchscreen were before it. In his Critical Path podcast Wednesday (Back to MOREPhilip Elmer-DeWitt - Nov 4, 2011 8:08 AM ET
An analyst may have put her finger on the conversational interface's killer app
In a note to clients issued Friday, Cross Research's Shannon Cross pivots from the Steve Jobs eulogies to take a closer look at Siri, the natural language interface that Apple (AAPL) unveiled the day before he died.
In particular, she singles out an application that wasn't in Scott Forstall's demos or Apple's slick promotional video, but which fits perfectly MOREPhilip Elmer-DeWitt - Oct 9, 2011 8:02 AM ET
Apple's Newton handwriting recognition system was undone by a comic strip
The signature feature on the iPhone 4S that Apple (AAPL) unveiled Tuesday is Siri, a so-called intelligent assistant that depends on a lot of fiendishly complex artificial intelligence systems -- voice recognition, understanding context and natural language -- working flawlessly a high percentage of the time.
Few companies know better than Apple the risks of counting on applied AI to sell consumer MOREPhilip Elmer-DeWitt - Oct 5, 2011 10:30 AM ET
A computer you can chat with. Hmm. Where have we seen this before?
It was hard for veteran tech reporters to watch Scott Forstall's demonstration of the iPhone 4S's new Siri "intelligent assistant" system (see here, starting at the 73-minute mark) without recalling one of Apple's (AAPL) most embarrassing episodes from the John Sculley era.
Sculley, whose previous job had been, in Steve Jobs' unkind words, selling sugar water for Pepsi, was MOREPhilip Elmer-DeWitt - Oct 5, 2011 10:19 AM ET
It seems the company's best days are behind it, or that Facebook is eating its lunch. But maybe Google is so far ahead of the game that we're the ones who need to catch up.
By Kevin Kelleher, contributor
What is going on with Google? Sure, today President Obama will be meeting with Eric Schmidt at a tech executive meet-up in Silicon Valley, but he almost seems an also-ran compared to the attention MOREFeb 17, 2011 1:20 PM ET
Watch it this afternoon, when an IBM supercomputer takes on two human champions
It's the biggest man vs. machine confrontation since Deep Blue humiliated Gary Kasparov on the chess board in 1997.
Over the course of three episodes -- February 14, 15 and 16 -- an IBM (IBM) supercomputer named Watson will challenge Jeopardy!'s most successful human contestants.
We're not privy to the outcome, but based on the level of pre-publicity generated by MOREPhilip Elmer-DeWitt - Feb 14, 2011 1:33 PM ET
The Semantic Web promises to make data and applications smarter.
By James Hendler, assistant dean for information technology, Rensselaer Polytechnic Institute
What do web giant Google (GOOG), the New York Times (NYT), the pharmaceutical leader UCB, and web startups Garlik and Bintro have in common?
They are among the approximately two hundred companies that have announced, this month alone, details of how they will be enhancing their businesses by using the emerging technology MOREOct 2, 2009 10:00 AM ET
|Yahoo to buy Tumblr for $1.1 billion: Report|
|Stocks on a roll: Yahoo, Microsoft stoke appetite|
|The Winklevoss twins are Bitcoin bulls|
|My very cheap day|
|Bernanke's advice for college grads|