The newest wave of interest in Artificial Intelligence is different from the two others I have experienced. In the mid 1970’s, I, and many others became enamored with the idea that we might actually be able to build tools that could think for themselves. Imagine, I thought…” a drill that knows where to drill and can get up to do it by itself while I sat sipping a Pina Colada on the beach!” This admittedly weird example may at first seem useless until you fill in the context behind it. The 1970’s ushered in a kind of microelectronics revolution of sorts in the United States. At that time I still owned a tube tester and needed it from time to time. Transistors were becoming “old school” and new integrated circuits began to be the rage. My Radio Shack soldering iron and stacks of old tube driven radios and televisions (from where my back stock of parts came) were quickly replaced by shelves bursting with those dark blue Integrated Circuit manuals from places like Hamilton Avnet and Texas Instruments.
It was then, in my early 20’s that I gave up. No longer was electronics something that an uneducated hobbyist could enjoy…it now took years of education to read through these manuals and put the puzzle pieces together in a way that made sense. The first IC’s were also so sensitive to being destroyed by static discharges, that to handle them, there were more wires on the handler than in the project! Lost forever was the joy of hearing your first static-filled blast from the local radio station as you made a receiver from one diode and an old pair of headphones. Lost was the satisfaction of experimenting with various sizes and configurations of resistors and capacitors, stuck through large flat yellow pieces of plastic cut from the side of a prestolyte antifreeze bottle, to achieve that much-needed, but simple result. No longer could you simply enjoy the sense of achievement from seeing a dead ensemble of plastic, glass and wires spring to life in your hands, if only for a few seconds before it caught fire sending you scurrying for a towel to extinguish it before it melted into a large yellow glob of something resembling the creature from the yellow lagoon.
Of course, you saved these failures, (if you could manage to scrape what was left off your workbench). Because they represented steps along your path to electronics enlightenment. I could always imagine sitting with my pipe at my workbench with several neighborhood kids staring wide-eyed at this mysterious yellow glob, handsomely exhibited at just the right angle inside a plastic display case. I could see myself saying wisely, “ That there kids is a super heterodyne receiver I built using a toothbrush, a clothesline and 3 old capacitors I found in a toaster in a pile of trash on the side of the road.” “WOW!!” they would all gasp as they stared in fascination up at the prickly yellow ,mass. Then I would deliver the coup de grace, “ Yep, I would say as I let out a trail of perfect smoke rings from my pipe. “I would stay up readin’ the Old Farmer’s Almanac so I could predict weather conditions to catch the right skips off the ionosphere… Listened in to people talking as far away as Russia one night! “No way Mr. B” they would exclaim as they ran home to tell their parents the exciting news. “Hey Mom, Dad, you know That old man down the street with all the junk in his yard…..”Who?” they would ask. “You know… the crazy old guy that stands out in the yard every Saturday in his underwear and screams at his lawn mower because it wont start.” “Oh. Yeah,” they would say. “what about him?” “He showed us all this neat radio stuff he built where he had to read a book to figure out the weather forecast so he could use it to listen to people in Russia!.” Their father would gaze over his computer screen long enough to ask, “Oh yeah, well why didn’t he just get the weather forecasts from the TV?” The kids would say, “That’s what we asked. He said the TV played the National Anthem at midnight and went completely off the air except for a test pattern that lasted all night until the next morning!!” The father would laugh and say, “Yeah…right! And you kids believed him? That old guy really is crazy! You kids stay away from him. He may be dangerous.”
Its amazing how that story digressed from one in which I wanted to cast myself as the hero to the truth so quickly! But I digress. We are talking about how to speak to a robot.
Computers changed the way we think about communications forever. Before computers suddenly became a necessity in every household, all communications were thought of as the sharing of information between like entities. Dogs spoke to other dogs in dog and not to people, Horses all talked only to other horses (except for Mr Ed), and so on. We also believed that a horse in France spoke the same “horse” as a horse reared on a farm in the United States. In other words, horses speak the same horse without accents or the need to explain what a new whinny might mean. People are different, however, because if they speak different people languages, they can actually learn someone else’s language to communicate with them.
Before computers came on the scene I never thought about the fact that different entities might have to do complex translations to communicate. For example, a computer “communicates using electrical impulses separated into patterns of 1’s and zeros. Digital communications is therefore a “representation” of how computers communicate with each other. People, on the other hand, use sounds made by pushing air over their vocal cords. We call the letters formed into words and sentences, Languages. Instead of ones and zeros, these human languages use semantic, and not digital rules as a representation to share the information that needs to be shared. If you can match the rules from one representation to another, you have a translator and Voila! You can understand each other So what about robots? If we believe that robots are just computers built to look like humans then we have already had about 50 years of innovation in digital communications to help us. It should be easy.
The fact however, is that the robots we envisioned in the 1950’s and beyond have suddenly changed in a very unexpected way. No longer is a robot just a a computer that looks like a human. Today’s robots are rapidly transforming into entities different from both humans and computers. Artificial Intelligence used to be called AI/ML (Artificial Intelligence / Machine learning) and there was very little difference in the two. The concept of a robot has morphed so rapidly, however in the last decade that ML is hardly ever spoken about and it is becoming clear that there are new and better ways for humans to communicate with new AI-driven robot entities other than language or digital electronics-based representations.
As one example let us imagine that there is a way to read thoughts in a human brain and download that energy to be understood by another human (Oh BTW there is). We would probably develop that capability by capturing electrical signals from different parts of the brain and using a complex data-based algorithm to decode it into language so we could understand it. Although it can be done, that is an extremely inefficient way to design a system. A Direct human brain to human brain interface would be much better faster and cheaper. The same is true for a brain to Robot “brain” interface. There are far better ways to process information on the robot side than old stale digital communications signals.
Having tickled your neurons a little, because you may never have had time to consider language representation in that depth, let me throw a bomb into the mix. Its easy to translate from one data or knowledge representation another. Knowledge is created when we relate ideas using context. (i.e., 1+ 2 = 3, There is A BRIDGE OVER THE RIVER, ETC.,) As long as all you want to do is share knowledge, that works great. However, if you want to generate RELEVANT & MEANINGFUL UNDERSTANDING from that knowledge you have a real problem. What many people fail to grasp is that knowledge is sterile and to turn knowledge into understanding requires relevance. In layman’s terms, “you can pass knowledge back and forth all day but it won’t do you any good.” In my next article, I will explain the differences between data, knowledge and understanding and how relevance can be achieved when Humans and AI systems communicate.