Random thoughts today…
I love to read Wired Magazine, but because I like to read it cover to cover and because I want to really digest some of the articles, I run a little behind on my reading. I currently have about 10 months of Wired Magazine on my “to read” stack.
I just finished an article in the March 2018 issue about Amazon’s attempts to train their Alexa to “chat” with people.
I have so many thoughts on that article and that goal, but first I also need to describe a Facebook video that appeared on my feed immediately after reading that article.
A little quirky habit of mine that allows me to sneak on to Facebook from time to time at work is that I watch most Facebook videos with no sound which can be pretty funny at times. So keep in mind, I watched this video with no narrative.
This video that I watched was a split screen video. On the left side of the screen, someone was demonstrating how to solve the math problem: 35 x 12. She began by breaking the problem down into components such as 30 + 5 x 10 + 2 and went on to illustrate how that worked by drawing boxes and diagrams and arrows and creating at least 5 other problems.
In the meantime, on the right side of the screen (see what they did there…the “right” side?), someone quickly did the math the “old” way then made a pot of coffee.
While Other Side person was still drawing boxes and explaining how to turn this math problem into many problems, Right Side person (probably someone about my age) solved the problem, went to the coffee maker, measured out water and coffee, hit the brew button, waited on the machine, poured a cup and was happily having their morning fix while Other Side person finally wrapped up all the little problems created from the original problems and happily produced the same answer Right Side person had calculated several minutes before the coffee was ready.
I do understand the new way of doing the math that has developed over the years is to help students learn to break down a problem into all its components to be able to explain that problem to a computer thus “train” the computer by programming it to solve the problem in logical steps.
I understand that.
But wait…keep that video in mind and turn with me to my rant about teaching machines to “chat” with us.
Amazon has determined that we want our machines to do more for us than turn on the lights and play music and select books for us. Amazon has decided that we want to be able to carry on a conversation with the machine, just a casual chat like we would have with a friend.
The article talks about dumping all kinds of information into Alexa’s machine mind, giving her(?) access to even more information and training her to access that information randomly to keep a conversation going for at least 20 minutes.
And they are almost there. In the admittedly one-year-old article I read, Amazon held a competition in which three groups won a total of over a million dollars in prizes to reward their ability to create a program that could allow a machine to talk to a human for nearly 20 minutes.
Stay with me for just one more leg of this rant.
Consider this…in a world where Rod Serling’s Twilight Zone has made a dramatic comeback on TV, in a world where a simple math problem can take a page of paper and 10 minutes to solve, in a world where machines can casually chat with us for 20 minutes…in that world, are we training computers to think like people, or are computers teaching us to think like machines?