The Current Great Hype

The way you hear it told, Machine Learning, equivalently Artificial Intelligence currently is the ‘bestest’ thing ever to happen to mankind — a new capability that enables us to outdo all of man’s historical accomplishments. Not only will machine learning help us create a better world, but also that we will arrive at the dubious distinction of being the only specie on earth to not any longer have to think or be thoughtful. You see, Artificial Intelligence will do it all for us, write novels for us, seed our land with corn, teach us how to improve our sex lives, fix our children’s despair, fix homelessness in California, end the war between Israel and Hamas, help Zelensky arrive at an improved valuation of the life of each Ukrainian, reduce the time to the choice of a Speaker of the House in the USA from two weeks to four working days, and on and on it goes.

The truth of the matter? The scientific evidence shows that if you want to perform any inferential analysis, typically, that old traditional econometric workhorse— simple regressions using either of Ordinary (OLS) or Generalized Least Squares (GLS) will outperform Machine Learning. The evidence is furnished in studies, such as, Masini, Medeiros, and Mendes (2023), Makridakis, Spiliotis, and Assimakopoulos (2018), Ahmed et al. (2010), Gilliland (2020), and Coulombe et al. (2022).

The Scientific Evidence suffices to…