A few years back I attended a conference at Stanford and watched a presentation by some of the folks at Numenta.  It was very impressive, and discussed the architecture of the hierarchical temporal memory (HTM) system.  There were several other presenters that discussed the learning of how the brain stores patterns, and does pattern matching.

What I really like is the thinking that dovetails with other articles and books that I have read, and how the brain – from childhood – constructs a model of the universe, and we the interact with the world using that model as a base reference for predictions and “next moves.”

Numenta has now released their Numenta Research Release.  It is now at the stages for researchers to begin to look at and experiment.  Of course it’s not just download and run … in Jeff’s own words:

We have found that it takes some developers months to get familiar with the
concepts, and maybe a year to get commercially useful results. It also is
possible that your first attempts at using HTM will be valuable for learning,
but not valuable for commercial applications. The learning curve is similar to
the learning curve for writing computer software. It takes anywhere from several
months to a year to become proficient in writing computer software; HTMs have
similar complexity.

The best part that I like about these theories and examples of possible brain models, is that we can learn a lot about ourselves from these.  How we work.  If the brain creates a model, and that is then used to interact with the world, then that model begins to define the limits of our interactions.  The funny part, is that most people fail to distinguish that it is simply a model … not the “truth”, and not fixed.  And so quite often their own model of the universe constrains their behaviors and actions, limiting what they can do.  I took a number of courses from Landmark Education
that focus on this specific area … learning to distinguish that model, and to then learn to evolve and update that model.  It is amazing experience each time I distinguish something new about how flawed my model of the universe is.

Even with the understandable doubts expressed by the author of this article … I am impressed with Jeff and his work, and believe that he is on the right track.

Quest for the Thinking Machine. Jeff Hawkins created the Palm Pilot and the Treo. Now he says he’s got the ultimate invention: software that mimics the human brain. By Evan Ratliff from Wired magazine. [Wired News: Top Stories]

My friend Dave Cline sent me another link today about Mechanical Turk. It’s a story from the MIT Technology Review about Pennies for Web Jobs. To me this is the type of article that backs some of of my theories on the coming Singularity.

From the article, I really liked this quote:

Not only did participants supply the necessary answers, but they did so
“outstandingly fast,” according to Cabrera, allowing Amazon to use the
photographs in its search results. “This is the tip of the iceberg, but you can
see how it enables ‘massively parallel’ human computing,” he said.

When I last met Vernor Vinge I spoke with him about my theories on how
to measure the presence of the Singularity. I proposed that we
might look to create a metric based on how many people are performing
machine driven work. Mechanical Turk is a very good example of
this … and yes it is simple … but there are people and systems
putting “work” into a large database, and there are people who are
executing on queues of tasks for money. To me, this really isn’t
that different from the little mouse pressing a bar for a piece of food.

A more complex example are the drivers for UPS and Fedex. Their
entire day is coordinated by massively complex computer systems that
manage the thousands of drivers all over the world. From the
beginning to end of the day they are simply following the directions of
computer systems that are managing a process far too complex for humans
to direct anymore. In fact, the computer systems are managing the
flow of packages and simply using humans as one of the components in that management system.

Vernor talks about a hard takeoff, and a soft takeoff of the Singularity. I will argue that we are already accelerating on our way in a soft takeoff.