Computer technology has become more and more powerful over
the last few decades. Yet still there’s a clear margin - actually a gap -
between Computers’ (artificial) decision making ability when compared with human
intellect.
Filling out this gap is one of the most important problems that needs to be addressed, because this digitalizing world requires devices and
applications that human can use more comfortably and effectively. Truth is, to make the
human feels comfortable, the ‘dumb’ devices or applications should anticipate what
he requires and consider his preferences. Therefore, artificial decision making
algorithms should at least resemble (if not match up with) human intelligence.
It’s true that Computer Scientists day by day progress in
Artificial Intelligence (AI), Artificial Neural Networks (ANN) and other related theories. However,
as it seems it’ll at least take few more decades (or even more) before logical
decision making, comparison or any other (currently human specific)
capabilities are to be successfully simulate by technology.
OK. But should we wait that much longer to get done real complex processing by machines? Is there any other path that we can follow, apart from waiting till artificial intelligence catch up real world needs?
Yes! There is an evolving branch of Computer Science called
‘Human Computation’. Human computation is basically about bringing the real
human intelligence into solving the problems that current artificial
intelligence algorithms are unable to solve. Technically, its outsourcing the
computations to human which are out of the scope of AI
algorithms.
A one good example is ‘reCAPTCHA’ project. You know what is
a ‘CAPTCHA’ – it is a program that can decide (actually, guess with high
probability of accuracy) whether its user is a human or a computer. reCAPTCHA,
in one aspect do the same thing – grade a user by letting him to complete a
test. In the other hand it helps to digitize books, newspapers and old time
radio shows.
Converting old time physical books in to digital material through OCR may lead to error prone results (again, because those artificial character recognition algorithms are not ‘human like’ intelligent). Currently available computer vision applications could not identify every letter, symbol taken from the scanned images of old books. But in reCAPTCHA, it is actually a human that will be converting the scanned words in to digital text. But these human operators are not doing this as a community service, they doing this because this used as a mechanism to filter human users from programmed bots.
Converting old time physical books in to digital material through OCR may lead to error prone results (again, because those artificial character recognition algorithms are not ‘human like’ intelligent). Currently available computer vision applications could not identify every letter, symbol taken from the scanned images of old books. But in reCAPTCHA, it is actually a human that will be converting the scanned words in to digital text. But these human operators are not doing this as a community service, they doing this because this used as a mechanism to filter human users from programmed bots.
By human computation, it may enable lots of things which otherwise
would be impossible. Such as identifying and tagging images over the internet which will ultimately enhance the images search results. Below is a video about using effective human effort in to
performing tasks that are not in the scope of artificial intelligence, but
which solving will undoubtedly help the human. This is a Google Tech Talk
delivered by Luis von Ahn who is an associate professor at Carnegie Mellon
University.
No comments:
Post a Comment
Had to enable word verification due to number of spam comments received. Sorry for the inconvenience caused.