2026-07-19 18:06 UTCIn-site rewrite6 min readUpdated: 2026-07-19 18:28 UTC

AI Requires a Labor Market Bailout

AI may eventually create more jobs than it destroys, but without a coordinated retraining effort, millions of lost jobs could unnecessarily become lost careers.

SourceHacker News AIAuthor: mooreds

AI may eventually create more jobs than it destroys, but without a coordinated retraining effort, millions of lost jobs could unnecessarily become lost careers.

July 14, 2026

min read

Technology

Education

Professional Development

‍

Image generated by DALL-E

‍

Historically, the job market’s demand for skills has evolved at a pace allowing the labor market to keep up. That began to shift in recent decades and is now woefully untrue. The rapid pace of change in workplace skills will not be met by private companies alone, which generally lack the means or incentive to upskill employees accordingly.

‍

Evolutionary Shock

I specifically used the word evolution since the analogy is apt. Over time, species adapt as the world itself evolves. As both their predators and prey change, so must they. But abrupt environmental shocks can produce changes so large and rapid that species cannot adapt quickly enough. The Cretaceous–Paleogene (K–Pg) extinction event, better known as the asteroid impact that wiped out the dinosaurs, transformed the environment faster than many species could adapt.

I first encountered this during the Great Recession. We saw millions of jobs lost in that period. In many prior recessions, layoffs were due to the cyclical nature of the business cycle. As the economy came back, so would those jobs. This time, however, many of the jobs disappeared permanently. We saw this in industries like finance, media, and marketing. (Living in New York City, I had a front-row seat to their elimination.) For a twenty-four-year-old, it was a speed bump in their career; for a fifty-four-year-old, it was a derailment.

I’ve written and spoken many times about prior changes in the job market due to technology in articles like “How to Know If Your Job Is Safe from AI — Part 1: What History Shows Us About Job Loss and Job Growth” and “No, AI Isn’t Going to Kill You, but It Will Cause Social Unrest — Part 1.” Those past changes have typically been widespread but slow, as with farming becoming mechanized over decades, or limited in size and fast, such as the replacement of elevator operators. The speed and breadth of change are well documented elsewhere, so I won’t cover them in depth. Research from the nonprofit METR (Model Evaluation and Threat Research) shows that AI’s task-completion horizon is growing exponentially, historically doubling roughly every seven months. Research by the World Economic Forum cited later in this article speaks to the widespread impact of AI on labor across multiple industries.

‍

The Buy-versus-Build Labor Problem

Today, with AI improving at unprecedented speed and finding applications across multiple industries and functional areas, tens of millions of jobs are at risk. I do believe that, in the long term, there will be new jobs. I also believe that, in the short term, there will be a shortage of the skills needed for the new jobs until more recent college graduates who have grown up developing these new skills enter the workforce (see “Where to Put the X in the Age of Generative AI”).

The question is: how do we close that gap? With this massive, widespread labor disruption, existing employees, like the dinosaurs, can’t change fast enough. Most mid-career professionals with a mortgage and kids can’t put their careers on hold for a year or two to retrain.

Companies won’t provide enough of that retraining. Companies face a buy-versus-build decision. Do you build and maintain your own custom in-house version of a product, or buy an off-the-shelf version that may not be 100% perfect but is sufficient? The same applies to labor: Do you train your own workforce or simply hire people who already have the skills you need?

Here’s the reality. According to 2022 Census Bureau data, nearly 90% of U.S. employer firms have fewer than 20 employees, and another 10% have between 20 and 499. These smaller firms collectively employ about 46% of private-sector workers. (Source: “Data by Enterprise Employment Size,” 2022 SUSB Annual Data Tables by Establishment Industry)

Smaller companies lack the scale to maintain sophisticated HR and learning-and-development operations capable of anticipating which skills will matter and teaching them effectively. Large employers have greater capacity, but capacity does not necessarily create an incentive to retrain workers when they can hire the skills they need instead. When companies need people with new skills, they are more likely to hire for those skills. This will lead to more companies competing for a limited talent pool, pushing up wages for in-demand workers while demand for workers with outdated skills falls.

It could be argued that employees should be responsible for upskilling themselves. It’s a very valid point. People should also be responsible for staying healthy as well as managing their finances to prepare for retirement and periods of unemployment. There’s a reason the government runs programs like unemployment insurance and Social Security and why most countries have national healthcare. People often struggle with long-term planning and decision-making. Investing time and money today for an ROI tomorrow isn’t in the DNA of most Americans. Additionally, workers will need guidance assessing exactly which skills will matter. It was only a few years ago that everyone said prompt engineering was the next hot job. (Well, almost everyone, see “Prompt Engineering Jobs are a Mirage.”)

The net result will be a wider gap between the skill haves and have-nots of the future. This will further discourage in-house training in the short term, since employers risk up-and-away-skilling employees who can leverage these in-demand skills in a very competitive job market. Eventually, the educational system will catch up and future workers will have the necessary skills, but many in the current workforce will get left behind, creating a drag on the economy.

‍

The Solution

In 2009 New York City’s Economic Development Corporation recognized that these jobs weren’t coming back and that a large part of the city’s labor force faced long-term unemployment. It funded programs like FastTrac and JumpStart at the SUNY Levin Institute with support from the Department of Small Business Services and the Kauffman Foundation. The former was an entrepreneurship boot camp designed to help people start their own businesses. The latter was an intensive retraining program that placed experienced professionals in internships with startups. While the idea of a fifty-year-old intern may seem odd, there’s no reason society shouldn’t be doing more of it. (Disclaimer: I was an instructor and advisor for JumpStart.)

New York City’s 2020 report, “Mayor’s Office for Economic Opportunity Independent Evaluation: Return on Investment Analysis of Industry-Focused Job Training Programs,” found that the 10-year taxpayer return on each $1 invested ranged from $0.36 to $3.68, averaging $2.02. A RAND Corporation study of a New Orleans job training program found its participants had, on average, 25% higher earnings than the control group. While results vary, these types of programs have generally been found to provide a net benefit.

Government bailouts are used to stimulate the economy in times of recession. We’re going to need a similar program on a national scale, even if locally executed, to bail out the labor market. It may be done by the government directly (federal, state, or local), through nonprofits, or through public-private partnerships. Any solution, however, will require the leadership and financial backing of the government.

To be clear, we’re not trying to hold back a tsunami but to ride the wave. For example, attempts to save coal jobs fly in the face of technological progress, while proposals to retrain coal miners for jobs in growing industries benefit from the tailwind of change. These programs must provide training for skills that are in growing demand. As you know from prior articles, I also believe that developing soft skills matters, even more so in the age of AI, alongside the technical skills required for a particular role. It may involve a combination of classroom training, mentorship, internships, support for new businesses, and financial relief, including tax breaks. While I believe this effort must be driven nationally, local execution is key. Communities will know best what their local needs are, although demand is often interstate, especially for online work.

This isn’t a must, of course. Financial stimulus helped turn around the Great Recession. Prior recessions and depressions had less government intervention and took much longer to recover from. Even without government action, the labor market will eventually adjust to AI. But why prolong the suffering when we have proven tools to reduce the disruption?

‍

The World Economic Forum’s 2025 Future of Jobs Report estimated that AI and other trends could displace up to 92 million jobs and create 170 million new roles by 2030, and that “39% of existing skill sets [will be] transformed or rendered obsolete by 2030.” (source) I am a short-term p(doom)er and long-term p(boom)er. That is, we will have massive labor disruption in the short term (two to seven years from today) and strong job and economic growth after. Exactly when, how long, and how extensive that short-term doom will be depends on a number of factors, including government intervention. My main issue is that while technological progress has been great for humanity overall, many of the workers displaced by prior technologies (farmhands during the Industrial Revolution, textile workers after the introduction of power looms, and tollbooth collectors displaced by electronic tolling) get left behind. I’m all for progress, and that sometimes means eliminating jobs. It doesn’t have to mean ending careers.

‍

Mark A. Herschberg

Back to blog posts