My latest feature in Wired is just out (on the newsstands at the moment; not yet online but I’ll post it when it is online here). It’s a profile of Numenta, a new company founded by Jeff Hawkins, the inventor of the Palm Pilot and Treo. Hawkins has been studying neuroscience on his own for the last couple of decades, and co-wrote a book about the fundamentals of intelligence and the human cortex, called On Intelligence, back in 2005. Numenta is his attempt to put the ideas from the book into practice, and build a new kind of artificial intelligence technology. It looks something like this:
Make sense? Well, hopefully the story explains in a somewhat intelligible fashion. I’m hoping to elaborate on a few things in the article here over the next week — there are some tricky issues in discussing any technology at such an early stage — but in the meantime feel free to offer opinions in the comments, or send me an email at the address to the right.
Posted at 4:46 pm | Filed under Recent stories, Technology, Wired |
Comments
5 Responses to “Software on the brain”
Leave a Reply
With recent articles in Business 2.0 and now Wired Numenta is aparently getting ready to go more public with their technology and open it up to a wider array of partners for experimentation.
Their approach has the potential to become a new way of building software. It might enable us to come up with ways of dealing with certain computing problems that are orders of magnitude better than we have now. It may also allow us to deal with entirely new classes of problems.
Such step changes (like the PC and Internet revolutions before) drive the entire industry forward and make it an exciting and vital place to work.
I look forward to reading the Wired article.
–
Fergus
Evan,
Have you ever read Roger Penrose’s The Emperor’s New Mind? In it he details the reasons he belives the human mind cannot be a turing machine, regardless of complexity. If he is right, then attempting to develop an algorithm to replicate the human mind is akin to trying to build a perpetual motion machine.
Peter,
I haven’t read The Emperor’s New Mind, although I have heard just a thumbnail sketch of his theories. My understanding is that Hawkins’ perspective, as outlined in On Intelligence, tries to offer a counter to theories like that of Penrose, although perhaps not Penrose specifically. In a larger sense, my guess is that Hawkins would argue that he is not trying to replicate the human mind in its entirety–since of course the human mind includes features (particularly those of the hippocampus) that will not be captured by HTMs–but that he is trying to capture the purely cognitive aspects that are involved in processing sensory input and making predictions.
But not having read Penrose, it could easily be that he offers a counter to that approach as well. I think there are certainly multiple schools of thought that cast doubts on what Hawkins is trying to do, but the good news is that he will either get scalable results in the coming years or not. So the proof will be there, or it won’t.
Your article is very well written and I love his ideas.
Great article, glad to see some coverage of Jeff Hawin’s work in Wired. I read the book, On Intilligence, loved it, and his ideas have inspired me to research the information he talks about, such as Bayesian inference. Probability, the logic of science. Check out Autonomy.com, they use Bayesian inference. The whole idea of being able for a system to “sense” a pattern, and that it is best for the system to be allowed to run on it’s own, and learn patterns. Ray Kurzweil did that with his work in speech recognition. Jeff Hawkins’ video at Mit, http://mitworld.mit.edu/video/316/, is what led me to read his book, and the funny thing is, I didn’t know he is the one that designed my Palm PDA!