Today, big data seems to be the main event. With the amount of data in our world growing at exponential rates, the question is “how do we make money off of that?” The answer is to “be more data-driven”. To be sure, this is unquestionably a good thing. Without data our decisions rely on intuition, and intuition is riddled with bias.

If you take the data-driven approach to the limit, then you run into problems as well. It can be tempting to hail the output of a machine learning model as truly unbiased conclusions, but this is also false. Machine learning models (which are built by people) have the same bias as your human ability to make decisions, but the computer lacks the common sense to filter out silly or dangerous ideas.

Economics seems to suffer from this problem as well. As a research field, there is a strong emphasis on developing robust statistical models that explain human decision making and policy effects. In particular, macroeconomics – an academic study of the most complex system of all: our global economy – seems to rely on mathematical models as a tool for making conclusions. A Bloomberg article covered Paul Romer, an economist who recently spoke out against the “mathiness” of his profession, and some of the backlash he faced from his peers.

Romer says: “Essentially, [economists’] belief was that math could tell you the deep secrets of the universe.” This sums up how I view the furore over big data and machine learning. Using data to understand a problem is good; relying on data to decide for you is harmful. Macroeconomists, whose roles include knowing how to grow the economy and prevent recession, can disproportionately impact millions of people’s lives. I’m hoping that we as a society eventually find the right balance of mathiness and human intelligence.

Unless the AI beat us to it.