Billionaires John Arnold and Mark Cuban have weighed in recently with ideas ranging from taxing labor at a lower rate than capital and taxes on AI-specific features like tokens and compute. The Economist’s recent coverage of the issue proposed a new fund to offset the costs of the economic transition for those impacted by AI tools. Sam Altman and Vinod Khosla have called for drastic tax cuts on workers, while others like Sen. Elizabeth Warren have made the case for wealth taxes in response to AI’s rise.

Underlying these recommendations is a sense that AI will fundamentally alter the economy. Workers could be displaced. Resources could be strained. And some use these risks to argue that tax policy also needs to change.

We are skeptical.

AI may be a transformative technology, but that is not a good justification for throwing core principles of tax policy out the window, certainly not based on what at this point is still pure speculation about the future of the labor market. Good tax policy involved simple rules, low rates, broad bases, and avoiding penalties for investment before and after the invention of the internal combustion engine, atomic bomb, and personal computer. The same goes for AI.

Labor-saving technologies have transformed work and life in the last century even while the shares of national net income that accrue to workers and capital owners have stayed roughly stable. The US labor market is generally more dynamic than many might think if one focuses solely on layoff announcements. In 2025, there were roughly 63 million hires across the US economy and 63 million separations like quits, layoffs, and retirements.

Major labor market disruption is a risk, but labor-saving technology does not automatically require unemployment. It could, instead, mean a significant increase in leisure: time spent outside of work with family, friends and neighbors. A transition for many to a four-day work week (without a corresponding 20 percent reduction in compensation) is possible in some sectors.

People have a right to be concerned, but those concerns should not lead to bad fiscal policy.

If past economic transformation changed tax policy historically, it has been in the direction of broader tax bases.

In the early stages of the Industrial Revolution, the United States relied primarily on tariffs and excise taxes on specific goods. These revenue options were preferred in large part due to their ease of collection in a still-modernizing economy: it was much easier for the federal government to collect taxes only at ports and distilleries than from each individual or business. Technological change made taxing broader bases, like income or consumption, more possible.

Specific taxes on tokens or compute would simply be counterproductive, penalizing the adoption of new technology. But they would also constitute a reversal of the wider trend for broader tax bases that historical improvements to technology have enabled. Stocks of AI companies have seen a dramatic increase in their value, but that value does not reflect taxable profits—yet. But if, or when, investors sell shares, those sales could generate taxable capital gains.

Additionally, property taxes didn’t suddenly cease to exist when data centers came on the scene. It might surprise some to learn that in 2025, Loudoun County in Virginia was able to lower tax burdens on residents partially due to booming data center activity (and related property tax revenue). But, policymakers should avoid creating special preferences or tax carveouts for data centers beyond what one would expect for any business.

Higher than expected tax revenue from capital gains, profits, property, or other existing taxes should be channeled to prudent ends. Federal deficits are high as far as the eye can see and states and localities also face budget pressures.

The principle of simple rules not targeting a specific industry or source of social change can also be found on the spending side of the equation.

Some say, if there is to be labor market disruption, shouldn’t there be a fund that is financed with the wealth created by AI? Couldn’t the proceeds from that fund be used to offset the costs faced by individuals whose livelihoods are disrupted by AI?

The federal government has experience in this area.

Trade Adjustment Assistance, or TAA, is a program designed specifically for helping workers disrupted by trade. However, TAA has long suffered with low uptake rates—few eligible people actually accessing benefits.

There are a few reasons for low uptake, but the most straightforward one is it’s tough to say trade policy caused a particular layoff. The same lack of clarity is already here regarding alleged AI-related layoffs. While we have seen large announcements of AI-driven layoffs, they have often come from businesses in some form of financial distress or that overhired in 2021-2022.

An AI-specific adjustment program for workers would risk falling into the same pitfalls as TAA. The alternative: fixing the existing unemployment insurance system. While it can be improved, and benefit uptake is not perfect, it performs a lot better than TAA because it is a broader policy.

AI may be one of the most exciting and frightening topics in public policy today. But neither excitement nor fear means that longstanding principles of sound tax policy have suddenly expired.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.