Good Morning Ted and Jody:
Yesterday afternoon, I send you a “forward” about how the digital world is changing so fast: the demise of Kodak to the rise of HUGHE companies that own no property such as Air B and B and Uber. How computers are getting so smart one was able to beat a computer at GO. Granted computers and digital software driven things are taking over. However, while Watson can win at Jeopardy and be more accurate than Lawyers, software cannot yet dance nor cook breakfast or more properly drive robots to dance or cook breakfast.
Computer software, while capable of learning, has been modeled or designed for goal oriented objectives: winning a game, meetings specific objectives, accuracy (getting the bolt in the right hole on the assembly line), consistently following a template. But moving one’s feet to music in coordination with another human being but not always in mirror behavior for the pointless outcome of having a good time (and getting some legitimate physical contact with another human being) but not getting to a specific spot on the dance floor is not yet and artificial intelligence achievable feat. Neither, is cooking breakfast so that everything always finishes in an appropriate time to have the eggs, bacon, hash browns and toast done in such as manner as to be both palatable while everything is of a consistent warmth and acceptable presentation. Yes, artificial intelligence programs, or even plain old fashioned software, can get the egg cooked. However, juggling all four actives at the same time and coming out with a product one would actually want to sit down to and enjoy is beyond the state of the discipline at present.
You may remember that over twenty-five years ago I took some coursework in computer science at Central Michigan University and Michigan State University. At Michigan State I took an artificial intelligence course. In that context of that course and a machine language theory course, I attempted to design a small piece of software that would learn. I failed to achieve the design. The problem then was similar to the problem now. I was able to get my software to repeatedly do something. However, I based my program on changing what it did based on reinforcement. When it did something I wanted it to do the theory was to reward it—reinforce the behavior. It didn’t work. You might say, “Well, Ed, you should have added some punishment so it would avoid doing what you didn’t want it to do.” Well, I couldn’t figure out how to reward it. So how was I going to tackle punishment. You reward people with praise, M&Ms, and the like. What in the hell do you give a software program that it would like?
At the time, I was attempting to do what 99.9% of artificial intelligence researchers and developers have been doing since John McCarthy invented the term “artificial intelligence”—design and write goal oriented software. However, learning in human beings may be goal oriented or may not. In any event, human behavior is largely learned through repletion accompanied with rewards (“Attaboy”) and punishments (“You blew that Bob”). But then some things human being learn by seeing it done or doing it once (I’ll leave the examples to your imagination and memory).
So, I suggest there will be a market for human beings to dance and cook breakfast and do all those other non-goal driven things we do with each other. Hence the forward is not as bleak as one might first imagine. Then too, someone has to pay for all this computerization. That is where we come in. At some point my proposal, https://democratizemoney.wordpress.com/ , or some other proposal for creating money in such a fashion that everyone has money so they can buy what they want from the computerized vendors of computerized manufactured and computerized grown goods and services and pay the dancers and breakfast cooks to keep dancing and cooking breakfast. So, you see, there is a rather positive side to this horror show of what the future will look like from the forward I sent yesterday.
Warmest regards, Ed