I have a cold. It is a very common cold. Nothing special. Just enough to make me want to sleep at 2 in the afternoon. We can land humans on the moon and yet we can’t cure the common cold.
Coincidentally, yesterday’s RadioLab podcast was indirectly about the common cold. The podcast was about complexity and the limits of human understanding. When we have complex systems, with many parts that interact and affect each other in complex ways, it is difficult for humans to make sense of the chaos. Despite its commonness, the common cold is one of these complex systems. The way that the cold virus interacts with the genes and proteins and other parts of the human body is extraordinarily complex. The human brain cannot make sense of such a large number of interactions. We have a limited capacity for understanding these kinds of interactions.
And so, according to RadioLab, we build robots. My sense is that these are artificial intelligence tools which map the interactions and attempt to create mathematical equations to model those interactions. And we have been successful in creating the equations that model those interactions. The problem is that we don’t understand the equations. We recognize that the equations are correct in the sense that they can predict the future. That is, given new data, they can predict the consequences on other variables in the system. These kinds of tests are common in neural networks, an artificial intelligence technique that attempts, in a limited sense, to model the human brain. One of the ways that we test whether a neural network has been created correctly is to measure its ability to generalize, to predict the results of new data. The problem with neural networks is that they can’t tell us WHY they predict what they predict. Instead, they create mysterious, almost mystical equations to predict the consequences of new data, to predict the future. But we can’t understand these equations. We know that the equations tell us that when one variable goes up another variable goes down by some fractional amount. But we don’t understand why. We cannot make sense of the equations. And so researchers are pushing our tools to explain themselves. To tell us why one variable goes up when another goes down or vice versa. Interesting.
And still, I have a cold. I know how this virus manifests itself in my body. But no one yet understands why. Or how to cure it.