Redefining Economics

As the field of Artificial Intelligence (AI) expands, governments and policy makers will need to make significant changes in their approach to modeling economics and in crafting policy guidelines for the new economy. The reason why this will be necessary is straightforward. Today’s economic models are based on theories originated in a time of scarcity. Productive capacity for many if not all goods were limited. Demand for products was relatively easy to identify and analyze. The equation was something along the lines of total income across all industries less savings and investments equaled total demand. Production of goods and services, with serious time lags, would catch up and then surpass demand and thus create economic boom and bust cycles. Later in economic history, we added credit expansion and market intermediaries that allowed for increased growth by leveraging the existing capital stock. 

In modern times, all of that information was codified, tracked, extrapolated, and implemented in the decision-making processes of businesses nearly instantaneously. Today, we have already entered the world of big data and soon, to accompany that, productive excess. Artificial Intelligence, Machine Learning, Cloud Computing, the Internet of Things, Alternative Energy and 3D Printing, to name a few recent technologies, will drastically change the economic landscape over the next few decades. Just as agriculture jobs once accounted for 90% of economic output, and the horse was the primary means of transport, new industries will gradually replace and expand our capabilities. In many ways, these substantive changes are not being correctly captured in today’s economic statistics and modeling. For example, the reference material of choice used to be the Encyclopedia Britannica which cost thousands of dollars for a set. Today, with access to the internet, you can instantaneously access virtually the full sum of human knowledge, effectively for free (as long as you verify the source). If today’s models capture this value at all, it is by calculating the monthly cost of internet access and the cost of the devices required to do so. Arguably, this is a cheaper, faster, and a broader knowledge base than the encyclopedia of old, but it is not valued as such. The implication is that today’s economic models and policies are behind the times. 

The other major impact to the economy is the changing composition of the work force. Not only do we have the issue of current demographics to contend with, but it is likely that many middle-tier clerical jobs may be reduced as AI and robot capabilities increase. In manufacturing, we already have companies whose robots are making robots. All this has important implications for policy related to educational training, retirement outlooks, and social welfare programs. 

The four primary areas that are key for policy makers and economists to focus on are:


1) Pending labor market changes and how to correctly measure productivity change.


2) Tax policy with respect to how to tax the productivity of robots as an offset to the changing tax base.

3) Governance in terms of how to address the use, ownership and privacy of data. 

4) Social Equity. The social equity discussion contemplates how to ensure that computer learning doesn’t inherit and embed existing bias into ongoing algorithmic learning. For example, as AI learns and mutates, if it sees that the CEO role is referred to in the masculine form 80% of the time, the robot might learn that bias. Alternatively, as the AI utilizes big data to learn about individual interests and preferences, they might then offer personalized pricing based on a potential future revenue stream. This could lead to further exclusion of lower income people, thus reinforcing rather than mitigating social bifurcation. 

All of these areas should be examined if we are going to be able to correctly redefine our economic models in the future.