My nephew vs ML
Eric Feldman
956

Your son is not learning like a machine learning algorithm. This is one of the biggest misconceptions that is leading to all the confusion surrounding AI and machine learning. Our brains DO NOT work like computers. In fact, we really don’t know how our brains learn. Comparing how an algorithm sorts pokemon cards to how your son does, is a completely irrelevant experiment. People/children learn in a much more integrative way. Like you said, your son was able to draw information from sources that weren’t part of the test and therefore make choices that the algorithm was unable to make because it’s source of information was limited to what you fed it. Children don’t only learn when we instruct them, children are like sponges, absorbing EVERYTHING they see hear, smell, taste, etc. Whether relevant to the specific task or not. That information is then filtered by the brain based on a relevance quotient that was predetermined by all of their previous experiences. Human beings don’t assess every piece of information available like an ML does. Human brains filter out massive amounts of information that they deem to be irrelevant focusing only on small pieces at one times: an ML learner always considers all of the data available to it and must incorporate that data. The two mechanisms aren’t even comparable.

So, good job building an ML algorithm that can beat your 4 year old son at pokemon card identification, I bet you helped build his confidence. You’ve demonstrated nothing, the comparison between your son and your algorithm is akin to a comparison between walking and an automobile. Both will get you to your destination, but other than having a similar outcome, they really have very little else in common. And then to declare the automobile the winner because it got there first, is completely missing the point about how and why humans learn.

Computer programs/algorithms are nothing like our brains, not even close. They will not evolve into “intelligence” they cannot become self aware. It’s just utter sci-fi nonsense. The human body is an organism that is comprised of billions of organisms, all of which affect the behaviours of the human. The bacteria inside your digestive tract actually communicates directly with your brain and can influence your behaviour. Did you add that to your ML algorithm? Perhaps your sun was hungry and his decision making was skewed by the fact that his body was desiring food and the game was an impediment to satiation so he wasn’t really paying attention. There are myriad factors that affect human decision making and learning, most of which we are unaware of. We’re hundreds if not thousands of years away from understanding how our brains work, yet for some reason, there are groups of people who believe that a coded program will be able to resolve the most elusive question of all humanity: “who am I?”

I call it modern day alchemy. As people of the past thought that they could turn less valuable metals into gold through chemical reactions. Some people today think that we can create “intelligence” or “awareness” with a coded program inside a box full of transistors, resistors and processors. Dream on if you like, but it just doesn’t make any sense.

Show your support

Clapping shows how much you appreciated Matthew Eisses’s story.