A generous friend regularly passes along the results of experiments he conducts by asking questions of various deep learning models, mostly involving investments he has made, or is thinking of making. The machines swiftly respond with answers whose quality varies. Because I am little involved in making investment decisions, I skim the results. But driving through Michigan last week to help open a family house for the summer, I participated in an experiment of my own.
As my daughter drove, I compared my recollection of point-to-point driving times with the instructions her phone was giving us, and with the arrival time that the machine was predicting. The latter didn’t change very much through the day. For quite a while I thought the machine was wildly optimistic. As we got closer to our destination, it became apparent that, just possibly, the artificial intelligence we were dealing with – by now the name I had given it was the Lying Machine – knew what it was talking about.
I was relieved, therefore, when, forty miles from our destination, the machine went haywire, directing us to turn onto an unfamiliar road in the Manistee National Forest, making me nearly certain it would miss its predicted arrival time. After a number of zigs and zags, all counterintuitive, we pulled up at our destination at the very minute it had first predicted nine hours before, and adjusted only slightly along the way. It turned out the next day that it had known about a newly installed round-about at the junction of US31 and M22 that had been slightly slowing traffic along our traditional route.
I have written this weekly newsletter for twenty years for two main reasons. The more prosaic is that it’s a means to an end: subscription income from the weekly keeps open an office while I finish what by now is a third book about developments in economics that took place in the years I covered the subject for The Boston Globe.
Most weeks I find some poetry in the letter-writing. What keeps me involved is the sense that I offer a parallax view of conventional wisdom in matters that, from previous experience, I know something about: economics, politics, newspapers, and America’s view of its place among nations. That means writing six days a week, and last week I took a short vacation. In three days of clearing, repairing, and visiting, we got the job done.
Two things happened along the way. I got carded buying a sixpack of beer. And we took a long walk along an usually expansive Lake Michigan beach.
It’s not often in your seventies that you must prove that you’re old enough to do anything, but you can’t buy beer at any age in Michigan without producing a driver’s license and letting them photograph it into a computer network. It had something to do with preventing vape sales to minors, I was told – a diminution of trust all around.
The twenty-meter sand beach was magnificent, especially remembering years in which waves lapped at the woods and tumbled an occasional house into the lake.
But where had the water gone? Into the atmosphere, hydrologists say, after an unusually warm winter during which little lake ice formed to slow evaporation.
And after that? Where the water falls out of the sky is for climate scientists to discover. That is a fundamentally trickier and more vexatious problem than having learned where it went.