Sunday, March 20, 2011

Symbiosis in the realm of the circuit



In my delving into the deeper world of technology, artificial intelligence (AI) and the world of human interaction, I am coming to some conclusions. First, humans are inherently flawed in their thinking process. Emotions, love, dreams-all of these things make us hopelessly and beautifully flawed. Second, computers and technology are endlessly evolving in a flawed linear path that makes their inability to think outside an algorithm beautifully flawed as well.


A year ago, 03, 2010, Wired Magazine published a short piece by Clive Thompson, "Clive Thomson on the Cyborg Advantage" in which he puts forth that the concept of cyborgs already exist. But first we need to start with what exactly is a cyborg. According to the dictionary on my mac it is:

cyborg |ˈsīˌbôrg|

noun

a fictional or hypothetical person whose physical abilities are extended beyond normal human limitations by mechanical elements built into the body.

ORIGIN 1960s: blend of cyber- and organism .

To go further, the origin of cyber comes from cybernetics:

cybernetics |ˌsībərˈnetiks|

plural noun [treated as sing. ]

the science of communications and automatic control systems in both machines and living things.

So a cybernetic organism (cyborg) is any system-whether organic, or mechanical, or both-that uses the science of communication. As Mr. Thompson asserted, we are already exhibiting cyborg traits in our use of technology to further advance our tasks to completion. Proof in point, as I was writing this post, I went to Google to search for my terms, validate my references and pull together the pieces I needed. I integrated my actions with my computer's to yield a better end result. What is missing from most assertions in the area of artificial intelligence is that we, humans are necessary to set this process in motion. I initiated this piece, not my computer. I knew what I needed to search for in order to assemble the necessary pieces to support my original idea. Had I not acted on the process, the computer would have sat idle.


The beauty of the flawed human thought process is in its ability to "create" that which rests outside of an algorithm. Look at the Netflix Prize winner . While Netflix offered a prize of $1M US, to any one who could who could create a predictive algorithm that would improve automated, movie recommendations accuracy by >10%, it took more than 7,000 teams and individuals over 3 years to create an algorithm using a database of over 1,000,000 movies to show an improvement in predictive modeling that yielded results based on a specified subset of data by a grand total of 10.06%. As a contrast, using Google, IMDB and an understanding of a person, we can get closer to >50% accuracy of predicting a movie a person will like over their having taken a blind guess. We could also flip a coin, but that's not the point. ( I should note here, I do not know the accuracy of correct prediction of the Netfilx algorithm either before or after the 10.06% improvement. Nor, do I have statistical data supporting general accuracy, so work with me here.)


The point is, until a computer recognizes a problem that needs solving, as well as having an understanding of abstract concepts like frustration or irony, it is outside the ability of technology to create the parameters that establish a framework to create the problem to which a dataset can be applied to reach a conclusion that resolves the original issue. Recognizing a problem through observation, creating the framework of the problem and then solving that same problem. That's true intelligence-artificial or organic.


While we look to such milestones as the day a machine passes the vaunted "Turin Test" , and takes home the $100K US Loebner Prize , that will only be the beginning. By the way, the current money is riding on 2029 as being the pivotal year for meeting the Turin Test milestone.


No, what will be the true turning point isn't the ability to fool a judge in to believing a computer is giving human answers, but for intelligent rationalization and observation to arrive at a conclusion. For instance, being able to observe and understand that someone you know well is having a good or a bad day. Sympathy, empathy and understanding. The ability to "read between the lines". For now, the best we can hope for in the foreseeable future is a blended alliance or process that combines the best of human instinct combined with computational analytics. As Garry Kasparov put it, "a weak human with a machine can be better than a strong human with a machine if the weak human has a better process."

No comments:

Post a Comment