The potential impact of Artificial Intelligence across all fields of study and aspects of life is arguably the most discussed topic in the 21st Century, to date. From the hours of online video content to the forum boards and real-life discussions on the subject, there exists an extreme societal concern regarding how these new technologies will change the world around us, whether our lives will be easier or if we will witness stricter competition in the workplace.
In the academic discipline of Economics, however, a notable subset of AI, namely “GenAI” or Generative AI, holds the potential to significantly advance economic research and learning while mitigating some of the challenges faced by economists and students of economics. By equipping the current and future Economists of the world with these sophisticated tools, we can expect a revolution in how research is conducted and the accuracy of economic forecasts. These new technologies will act as a catalyst for a revolution in Economics, resulting in the shedding of its old image as an outdated field of study, with an academic “old guard” resting upon their ivory tower, to a modern field of science, much akin to how we currently view Data Science.
Aside from the rather obvious and well established uses of GenAI, including OpenAI’s ChatGPT and Google’s Bard, which are currently finding use by condensing advanced material for economics students and assisting in research through condensing code and writing mathematical derivations, GenAI is being used in the creation of increasingly accurate forecasting models, particularly in regards to inflation, due to how it accumulates knowledge. Since GenerativeAI is in essence a form of advanced pattern recognition software, highly trained by hordes of data already existing on the internet, it has the ability to filter through information far more efficiently than humans ever could. ChatGPT, for instance, trains using data from books, wikipedia, and news articles.
Therefore, since ChatGPT, and GenerativeAI as a whole, operates on massive amounts of pre-existing human knowledge and modern online repositories, and has the ability to select an optimal answer to any question posed, albeit with some caveats. As GenAI essentially just spits out preexisting knowledge, it has its shortcomings, and especially struggles with logic puzzles.
Anecdotally speaking, ChatGPT seems to notably struggle with formal (sometimes called “mathematical”) logic notation and proofs, such as those found in philosophy and mathematics curricula. Returning to the discussion on more accurate forecasting, a working paper by policy advisors at the St. Louis Federal Reserve found Google’s PaLM, a GenAI similar to ChatGPT and Bard, was able to produce inflation forecasts with fewer errors than those produced by the Survey of Professional Forecasters, an association of highly trained economists who all hold advanced degrees in the subject. Examples of AI’s successes such as these serve as proof of the potential of the field of Economics to transform itself from a social science to a hard, evidence-based science, which is a must for the field if it wishes to maintain or increase credibility.
The introduction of AI to the field of Economics may also assist in the field’s de-politicization and biases, thus furthering the advancement of the field in becoming a hard science. Until fairly recently, economists, who conduct their research well within the operating bounds of their political biases, have fought the introduction of AI into their field, frequently positing that their heavily theoretical models are useful in analyzing markets, which they believe reflects decisions made by rational agents, despite the fact that there exists a large amount of disorder and irrationality amongst the economic agents which make up these markets. This is an almost obvious fact, especially given how many recent economic downturns across the world are a direct result of not recognizing our irrationality, such as the Dot-com bubble burst in the early 2000’s, which was a result of irrationality on the end of investors, who were sold on the novelty of the dot-com realm, without properly considering the viability of many of the early startups. By introducing heavy amounts of data to Economics, and the proper tools to analyze it, the field will inevitably exchange its present of politically-charged theories and outdated models for a future of objective answers, driven by AI.
The rise of Artificial Intelligence, particularly Generative AI, aka “GenAI”, presents transformative potential for the field of Economics. GenAI, exemplified by tools like ChatGPT and Google's Bard, is already streamlining research and enhancing learning by simplifying complex materials. Furthermore, its role in creating more accurate forecasting models, as evidenced by Google's PaLM outperforming traditional methods, signifies the potential for economics to evolve from a social science into a data-driven hard science. Additionally, the integration of AI can contribute to depoliticizing and mitigating biases in economic research, paving the way for more objective, evidence-based approaches and boosting the field's credibility and relevance.