Friday, July 19, 2013

Intelligence: artificial and natural

As I have written before, there are something on the order of 100 million blogs on the internet.  So, naturally, neither you nor I are going to read even 1% of them.  Way before you got very far in the project, your common sense and your perception of the world and your needs and interests would kick in and you would go do something else.


Google Reader was something of a hub into which I could put subscriptions ("feeds") when I came across a blog that seemed promising. For some reason, Google decided to discontinue that project.  They gave us plenty of warning. I looked up alternatives and quickly settled on Feedly, which seems definitely better than Reader was.  It says I am subscribed to 102 blogs, all free, of course.


I still continue the practice of having snippets of some of the blogs I follow appear beside my own blog posts on my blog web page.  I have not evaluated all my subscriptions lately to see if there are some I should remove from the web page of the blog and some I should add.


Today, I looked at Eric Barker's "Barking Up the Wrong Tree", a blog that often has good excerpts from insightful psychology books.  When I look at something in Feedly, I often click to go the web site for the whole article.  Once there, the links and ads may distract me or enhance my search and I get off into something else.  Somewhere, today, I came across a link to this PBS video on artificial intelligence, a subject that any researcher and any educator might find of interest.  One reason to pay attention to the subject is that work in it often highlights what seems to be an ordinary human ability that the scientists and engineers simply can't figure a way to duplicate.  Following a moving object with your eyes and understanding the sentence "To is not the same as too" may seem ordinary but they are complex and high-level.


Barker mentioned the book "Ungifted: Intelligence Redefined" by Scott Barry Kaufman.  The book has high ratings and Amazon says it is by a present day psychologist who was labeled in school at one time as needing special education. In the PBS video, scientists familiar with the work on simulating and attempting to equal or surpass human intelligence with machines admitted that much remains to be done.  They cite our vision, our use of natural language, our ability to manipulate objects and our common sense as especially difficult areas for artificial intelligence.  The video says that during the first few decades of attempts to build smart machines the general strategy was to create rules that the smart machines would follow.  Over time, it became clear that humans do so many things and change their approaches to problems and goals so often and so cleverly that unwieldy sets of rules would be needed.  Today, the work is proceeding more along the lines of trying to build machines that can learn the way we all did and do.



--
Bill
Main blog: Fear, Fun and Filoz
Main web site: Kirbyvariety


Popular Posts

Follow @olderkirby