AI’s Biggest Employment Impact May Be a Rising Skill Floor
Artificial intelligence’s biggest impact may be on how high it raises the minimum skill threshold for employment and not solely on the number of jobs that it eliminates.
Two forces could converge leading to high rates of unemployment. The first is increased technology-induced displacement, and the second is rising “skill floors,” by which I mean the minimum level of expertise and productivity a worker must demonstrate to justify being hired over an automated alternative. Here, AI may simultaneously eliminate accessible jobs while raising the skill threshold for those that remain.
Despite the potential for AI to create countless new employment opportunities, the time it takes new workers to develop the skills to work these positions may be too great to effectively transition workers. This could produce structural unemployment much faster than most institutions are preparing for, and rising standards for recent computer science graduates may be an early indicator we are headed in this direction.1
Pseudo-automation
Automation has long displaced workers in the labor market and is contributing to lower demand for many middle-skilled positions.2 I will not go into all the ways we are seeing AI displace labor, but I want to highlight a scenario I am becoming increasingly concerned with: AI’s ability to facilitate pseudo-automation. I use pseudo-automation to describe technologies or systems that enable labor to be pushed further down the supply chain, eliminating a paid position without fully automating the tasks away. For example, self-checkout lines shift tasks that were previously performed by employees to unpaid customers.3 The tasks still exist, but the paid position does not.
Because AI can be tailored to help in individualized situations, unlike the internet where documentation is inherently broad, AI can be used to supplant knowledge-based proficiency.4 So, it seems likely that AI programs will be created to cut out knowledge workers that were previously relied upon for tailoring their broad knowledge for individual situations. As the saying goes: “You don’t pay a plumber to bang on the pipe. You pay them to know where to bang.” If a hypothetical plumber.ai could walk people through simple plumbing repairs that required only a wrench, many plumbers could lose business from this technology — no robot plumber required. We may be seeing an early version of this pseudo-automation with vibe-coding and programs like Base44 cutting out programmers as end users are able to create their own unique software without needing technical expertise. At scale, we could see many workers get displaced as knowledge becomes less of an asset and tasks are shifted further down the supply chain.
Rising skill thresholds
The second force affecting the labor market is the rising skill threshold required to justify hiring employees. One early indicator of this trend is the disappearance of value-adding tasks that were previously performed by entry-level employees.5 Again, we see this effect on software engineers where AI seems to be contributing to a slowdown in entry-level hiring, while at the same time highly skilled developers remain extremely valuable and are earning higher salaries than ever.6 Although the impact is currently limited, these conditions may spread. Those more skilled than AI are rewarded, and those who cannot reach that bar are seeing fewer opportunities to use their knowledge.
The St. Louis Fed describes recent labor-market conditions as consistent with a “low-hire, low-fire” economy where companies are looking to hold on to their current employees without hiring new workers.7 Recent college graduates and young people may be struggling to find employment because, even though they may have developed some expertise from their education, they do not have the skills to outcompete AI or the globalized economy. However, those already employed are still able to justify their position. This may be from inertia, but it also may be because they have had time to develop their skills beyond what AI is currently capable of, making them remain valuable in the labor market.
If we continue to see a rise of displacement from technology, as seems possible through pseudo-automation, and the bar to justify employment continues to increase and begins affecting positions beyond entry-level roles, these could combine into a systemic problem that may lead to massive unemployment.
New jobs are not enough
The current leading rationale for why AI will not lead to systemic unemployment is that technology has largely led to the creation of new jobs either directly or indirectly.8For example, with the rise of the automobile, we saw increased jobs in automobile manufacturing, and we also saw increased jobs in related fields through gas stations, taxi drivers, and road construction.9 Those who lost their jobs in carriage maintenance could largely transition into one of these other roles with limited training.
Get Cole Scott’s stories in your inbox
Join Medium for free to get updates from this writer.
Yet this traditional backstop that prevented large-scale unemployment from technological displacement may be weakening. The newest positions created from new technologies have a higher skill floor as graduates with advanced degrees are 2.9 times more likely to be employed in new work than those with a high school degree, up from 1.3 in 1950.10 So even as technology creates more jobs than it displaces,11 we can still see large unemployment if workers cannot transition into these new positions. The difficulty of effective transitions is highlighted with the displacement of fossil fuel workers where only 0.7% transitioned into a clean energy job.12 Just because there are growing opportunities in one industry does not mean it is a viable landing space for displaced workers from another.
Although this situation is not inevitable, more people may have their jobs displaced from AI agents and pseudo-automation, and for many, the training needed to become valuable in the labor market again could leave them unemployed for significant periods of time. We should begin planning policy solutions now to prepare for the possibility of a burgeoning skill gap that I see as likely to accelerate in the coming years. In the coming months, I will look into apprenticeships, UBI, AI taxes, and many other potential solutions to see how we can keep the next generation of workers employed and thriving in this next phase of the American economy.
Like what you’re reading? Find me on Substack.
1 Rak, How to stay ahead of AI as an early-career engineer, IEEE Spectrum, Dec 25, 2025.
2 Autor, The Polarization of Job Opportunities in the U.S. Labor Market, Center for American Progress and the Hamilton Project, 2010
3 Moradi, Levy, Cheyre, Pseudo-Automation: How Labor-Offsetting Technologies Reconfigure Roles and Relationships in Frontline Retail Work, 2025.
4 Acemoglu, Kong, and Ozdaglar, AI, Human Cognition and Knowledge Collapse, 2026.
5 Leopold, How AI is reshaping the career ladder, and other trends in jobs and skills on Labour Day, World Economic Forum, April 2025.
6 Mahoney, McEntarfer, Wahal, Stanford Institute for Economic Policy Research, What Is Really Happening to Jobs? Separating AI Hype from Reality, 2026.
7 Rodgers and Kassens, How Shifts in Labor Supply and Demand Shape Outcomes for Young Workers, Federal Reserve Bank of St. Louis, June 30, 2026.
8 Autor, Why Are There Still So Many Jobs?, 2015.
9 Ibid.
10 Autor, Chin, Salomons, Seegmiller, What Makes New Work Different from More Work? 2026.
11 Autor, Why Are There Still So Many Jobs?, 2015.
12 EM Curtis, O’Kane, Park Workers and the Green-Energy Transition: Evidence from 300 Million Job Transitions, 2023.
