Working Up

by Dwayne Phillips

There is nothing wrong with mimicry. Let’s stop kidding ourselves; much of today’s machine learning is simple mimicry.

mimicry: noun, the action or art of imitating someone or something, typically in order to entertain or ridicule.

Let’s be candid here. Much of Machine Learning (ML)—a currently popular branch of Artificial Intelligence (AI)—is mimicry. Point to a million pictures and say, “These are giraffes, these are penguins, and these are snakes.” The system will be able to mimic you and correctly say, “These are giraffes, these are penguins, and these are snakes.”

That is a simple case of ML. There are more complicated ones, but those are more complicated types of the same thing. “I sat on a fill-in-the-blank.” What will I put in that blank? Probably not “nail” or “bumble bee.” Probably “chair” or “couch.” Simple mimicry.

Is that intelligence? Again, let’s be candid. The answer is, “No.” Is that useful? Well, in many situations it is quite useful and in many situations it enables a machine to do a rather mundane job. That is also useful.

The current ML branch of AI can be quite useful and profitable. Let’s stop there and not delve into calling it intelligent or sentient.