top of page
Writer's pictureThe Analyst

The AI-Effect

As opposed to our usual lighthearted and relatively short articles, my friend Josie gave us her Harvard essay for a competition. This essay takes on a more academic tone and covers quite a heavy subject of AI and its effect on the world. This is a critical topic and is crucial in today's world. I have also made a few short edits.

The AI-Effect

The sudden emergence of OpenAI’s ChatGPT has reignited the public’s interest in Artificial Intelligence (A.I.) and its implications for the future. As A.I. systems are progressively able to equal or even exceed human capabilities in a growing array of tasks, they are being integrated into various economic markets and their potential impact on the economy is becoming more and more apparent.

There is no doubt that A.I. benefits the economy; A.I. is expected to increase global GDP by 7% (or almost $7 trillion) and heavily contribute to overall productivity gains (lift productivity growth by 1.5 percentage points over a ten-year period). However, with such a short experience of A.I. so far, it is difficult to ascertain its precise effects on the economy; particularly the negative externalities of A.I. growth. This paper explores the potential negative effects of A.I. on the economy, and how one can prepare for its microeconomic and macroeconomic impacts.

Technological Unemployment:

Adam Smith’s Natural Law of Competition argues that competition forces people to make a better product for a lower price. The incredible profits generated by corporations over the last twenty years have largely been due to the impact of technology, which has reduced their costs of production; A.I. will accelerate this trend, but in doing so will also increase the unequal distribution of these gains. In his book “The Rise of the Robots: Technology and the Threat of Mass Unemployment,” Martin Ford describes Japan’s Kura sushi restaurant — a restaurant chain that has successfully developed an automation strategy. By replacing all manual labor with technology, Kura is able to price their sushi dishes at 100 yen (equivalent to around $1), whilst keeping a healthy operating margin — thus undercutting all its competitors and ensuring its success in a fairly saturated market. Unfortunately, this sharp decrease in marginal cost comes at a cost; although selling dishes at this price is increasing the company’s profits and a customer’s purchasing power in the restaurant, Kura has diminished the economic value of an employee to near-zero. Taking this example to its natural conclusion, reducing the economic value of any type of employee to zero will lead to mass unemployment.

Our economy has been no stranger to technological unemployment — this just echoes the cries of textile workers of the 1880s or switchboard operators of the 1970s. The main difference is the depth and speed of the potential threat A.I. poses to the labor market. COVID provided the catalyst for the increased rate of adoption, but this is being sustained by increasing shareholder pressure for returns. Today, around thirty percent of all tasks are done by machines — the rest are completed by manual labor. However, it is predicted by the World Economic Forum that by 2025 that the balance will change to a 50-50 composition of manual and automatic labor. Already in May of this year, 3900 of 80,089 job cuts that month were directly related to Artificial Intelligence. Thus, the ratio of human labor to automation is balancing out at a faster rate than it has ever done before; at a rate resulting in levels of unemployment that the economy is not prepared for.

The Robot-Effect:

Technology’s bull’s eye has always been low-wage jobs that consist of repetitive tasks and require little to no education or training. Wage declines among blue-collar workers whose tasks were replaced or degraded account for fifty to seventy percent of changes in U.S. wages since 1980 — automation being the primary driver in US income inequality over the past 40 years. However, A.I. is now targeting white-collar jobs in the services sector; for example translation services, data entry and paralegal work. For example, technology has evolved from replacing fast-food counter workers with large touchscreens, to threatening manager positions as A.I. systems are now able to complete basic operational functions. A well-known study by Carl Frey and Michael Osborne “The future of employment: How susceptible are jobs to computerisation?” states that almost half of jobs in the U.S. are at risk of being automated through computerization (including AI) over the next two decades.

This employment substitution could lead to greater wage disparity, due to soft skills not yet replaceable by AI being at a premium. In 2020, MIT professor Daron Acemoglu conducted a study in which he discovered that in the U.S., every robot added to the workforce per 1,000 workers resulted in wages decreasing by 0.42%. This wage disparity between jobs susceptible to being replaced by technology and those that are not will inevitably increase the economic inequality between high-income and low-income individuals.

In addition, this inequality will be exacerbated by the current inherent bias found in A.I. systems, due to the nature of the data sets upon which they are trained. For example, if one takes the data of all successful mortgage applicants over the last twenty years and then only uses this to extrapolate credit-worthiness — without any human oversight of evolving factors — it would mean that all these newly displaced workers wouldn’t even have a chance for upward mobility.

How should we approach fixing this social and economic cost?

Economist Maynard Keynes believed that the adoption of technology happened at a faster pace than the economy’s ability to adjust to its resulting unemployment; this, he argued, was one of the key factors in causing the Great Depression. Policymakers should heed Keynes’ warning and try to slow the effects of AI on the labor market. In order to blunt the impact this job displacement will have on the economy, we must slow down the integration of A.I. into the labor market. One way to do this is by imposing a Pigouvian tax. A Pigouvian tax helps solve negative externalities by implementing a tax equalling the social cost of the negative externality. This raises the price of the product, making people less inclined to purchase said product and decreasing the quantity demanded; thus, achieving the ideal equilibrium. Circa 2020, a potential “Robot Tax” was the topic of discussion amongst politicians and prominent figures including Bill Gates and NYC Mayor Bill de Blasio, where they argued that since payroll tax savings are one of the biggest incentives for employers to choose automated labor over manual labor, the Robot Tax should be equal to: (number of robots employed) X (average payroll tax paid by the employer per employee)

An “A.I.-Tax” would work in the same way. Setting an A.I. Tax would have the following two impacts:

1) Slowing down the diffusion of A.I. into the labor market

The goal of the A.I. Tax would be to disincentivize firms from replacing workers with robots or A.I. too quickly — allowing for equilibrium to be reached. Employment taxes on automated labor would be the same as human labor, thus employers would make neutral hiring decisions — at least in terms of tax-related spending. In addition to an explicit payment, this could take the form of forcing employers to take on the burden of retraining or finding the affected worker a new job. While A.I. is still equalling and not exceeding human capabilities at specific tasks, this will slow down the integration of A.I. into the labor market.

2) Makeup for federal revenue losses

Eighty percent of U.S. federal revenues derived from labor; a displacement of workers by A.I. would clearly require a source of revenue to gap that shortfall. An A.I. tax would compensate for this loss in revenue. As the advancement of A.I. can provide society with countless benefits, we should not halt its innovation. However, we should be aware of all potential consequences — such as mass unemployment and increased economic inequality — and policymakers should use all economic tools available to ensure the market is prepared. In other words, we should embrace A.I. with open arms, but keep a discerning eye on its ramifications.

“In short, it is only humanity which is capable of awe, which will also be capable of controlling the new potentials which we are opening for ourselves. We can be humble and live a good life with the aid of the machines, or we can be arrogant and die.” - Norbert Wiener for The New York Times, 1949

Essay Written by Josefina Calo, edited by Aurore Lebrun and Annika Bjerregaard

128 views1 comment

Recent Posts

See All

1件のコメント


ゲスト
3月18日

Very informative

いいね!
bottom of page