Federal Reserve Governor Michael Barr warned on November 11, 2025, that firms are already scaling back hiring because of artificial intelligence, a signal that the long-predicted collision between AI and the labor market is no longer theoretical. That same week, fresh research and government data confirmed that the effects are spreading well beyond Silicon Valley, touching occupations across the income spectrum. For workers, investors, and policymakers, the question has shifted from “will AI disrupt jobs?” to “how fast, and who absorbs the cost?”.
The Fed Connects AI to Hiring Slowdowns and Market Risk
In his November 2025 remarks, Governor Barr drew a direct line between generative AI adoption and three channels of economic uncertainty: productivity gains that remain hard to measure, a labor market where firms are scaling back hiring as AI handles more tasks, and financial stability risks tied to AI driven trading. Barr specifically flagged the possibility of increased market volatility, price manipulation, and even tacit collusion among AI-powered trading systems. These are not abstract concerns; they represent the Fed’s working framework for deciding how aggressively to adjust interest rates as AI reshapes the economy’s supply side, and they hint at a world where central bankers must read not just inflation data but also model architectures and training regimes.
Barr had laid the groundwork months earlier. In a February 2025 speech, he contrasted a gradualist path for AI adoption with a scenario in which the technology drives a more radical transformation of labor markets and productivity, explicitly tying generative systems to wages, employment levels, and inflation dynamics. The gap between those two paths is exactly what makes the Fed’s job harder: if AI boosts output without proportional job losses, rate cuts can proceed with less fear of overheating, but if it displaces workers faster than new roles emerge, the central bank faces a stagflation style puzzle where unemployment rises even as profit margins expand. For now, Barr’s message is that AI is no longer a distant tail risk. It is an active factor in monetary policy deliberations.
Government Data Already Shows Programmer Jobs Shrinking
The Bureau of Labor Statistics offers one of the clearest early signals that AI is altering hiring plans. In its Occupational Outlook Handbook, the agency’s entry for computer programmers projects a decline in employment from 2024 to 2034, explicitly attributing the drop to automation of routine coding and the spread of AI tools that allow fewer developers to do more work. For a profession once seen as a safe harbor in a tech-driven economy, the reversal is striking: the government’s baseline expectation is now that demand for traditional programming roles will contract even as software permeates more industries. Behind that headline forecast sit detailed tables and scenarios accessible through the BLS’s interactive research interface, where users can see how shrinking programmer headcounts are already baked into long-run models.
The programmer decline matters as a leading indicator because software development sits at the center of AI’s current capabilities. If the occupation most familiar with AI tools is losing headcount, the pattern is likely to spread outward into fields that are only beginning to integrate automation. An analysis by the OECD of regional labor markets in advanced economies found that substantial shares of workers hold jobs where tasks could be done in half the time with AI assistance, especially in knowledge-intensive services. That “half the time” framing is crucial: employers do not necessarily eliminate positions overnight, but as AI compresses the hours needed for core tasks, managers can freeze hiring, consolidate teams, or outsource work, producing a slow squeeze that standard unemployment metrics may miss until it shows up as stalled career ladders and rising underemployment.
Research Quantifies Who Gets Hit and How Fast
Two major research efforts help quantify how widely and how quickly AI could reshape work. A study by OpenAI/OpenResearch and University of Pennsylvania researchers, often cited under the label “GPTs are GPTs,” estimated the share of the U.S. workforce with at least 10 percent of their tasks exposed to large language models, along with the share of all tasks that could be sped up with LLM-based tools. The authors found that exposure is not confined to coders and data scientists; legal services, finance, and a broad swath of administrative and customer-facing occupations all contain large bundles of text-heavy tasks that models can already draft, summarize, or analyze. In many white-collar jobs, the study concluded, a majority of daily activities could be either directly automated or significantly accelerated by generative systems.
A separate paper by Anthropic-affiliated and external authors took a bottom-up approach, mapping millions of real-world Claude interactions onto O*NET tasks and occupations. That work confirmed heavy concentration in software and writing tasks, but it also revealed surprising breadth: a meaningful share of occupations already use AI for a substantial fraction of their work, from drafting marketing copy to generating financial reports or technical documentation. Crucially, the authors distinguished between augmentation, where AI helps workers produce more or higher-quality output, and automation, where it replaces human effort entirely. Their results suggest that, at least in the near term, augmentation dominates, but as tools improve, the share of tasks that tip into full automation grows, raising the risk that some roles will hollow out into thin layers of oversight wrapped around machine generated content.
From Productivity Hopes to Uneven Labor Market Pain
For businesses, the promise of AI is higher productivity and lower costs; for many workers, the reality is a creeping sense that their job descriptions are being rewritten without their consent. Early adopters report that generative tools can draft code, emails, and analyses in minutes, transforming workflows that once required hours of focused effort. In theory, this should free up employees to focus on higher-value, creative, or interpersonal tasks, and some firms are indeed redesigning roles to emphasize judgment and relationship-building over routine production. Yet the same efficiencies also make it easier for managers to justify hiring freezes or to restructure teams around a smaller core of highly skilled staff supported by AI, especially in sectors where demand is not growing fast enough to absorb the productivity gains.
Evidence from academic and policy research underscores that the pain will not be evenly distributed. A Brookings Institution analysis published in January 2026 argued that most exposure metrics ignore workers’ ability to adapt, emphasizing that those with limited transferable skills, weak local job markets, or barriers to retraining face the highest welfare costs if displaced. When AI erodes the value of specialized but narrow expertise—say, in routine legal drafting or standardized financial analysis—midcareer workers can find themselves competing with both younger, more tech-fluent colleagues and automated systems. Without targeted support, the gains from AI-enabled productivity could coexist with deep pockets of long-term unemployment, particularly in regions already struggling with industrial decline.
Policy Choices Will Shape Who Bears the Cost
Policymakers now confront a narrowing window to turn scattered pilot programs into a coherent response. Central banks, led by officials like Barr, are grappling with how to interpret productivity spikes and localized job losses in their models, wary of cutting rates too quickly if AI-driven efficiencies mask underlying weakness in labor demand. Fiscal authorities face a different challenge: how to finance and deliver large-scale training, career counseling, and income support without stifling innovation or locking in incumbent advantages. Traditional tools (tuition subsidies, community college programs, relocation grants) were designed for slower, sector-specific shifts, not for a wave of change that can reach legal clerks, copywriters, junior programmers, and back-office staff at the same time.
The policy menu is expanding to include ideas once considered fringe, from wage insurance for displaced midcareer professionals to portable benefits that follow workers across gigs and short-term contracts. There is growing interest in tying tax incentives or procurement preferences to demonstrable investments in worker upskilling, nudging firms to share more of the adjustment burden. At the same time, regulators are beginning to ask whether AI deployment in high-stakes domains—finance, health care, critical infrastructure—should come with explicit obligations to monitor labor impacts and report on how automation gains are translated into new roles. The core question is no longer whether AI will reshape the labor market, but whether the transition will be managed in a way that spreads opportunity or one that concentrates both economic and political power in the hands of a few AI-enabled winners.
More From The Daily Overview
*This article was researched with the help of AI, with human editors creating the final content.
Grant Mercer covers market dynamics, business trends, and the economic forces driving growth across industries. His analysis connects macro movements with real-world implications for investors, entrepreneurs, and professionals. Through his work at The Daily Overview, Grant helps readers understand how markets function and where opportunities may emerge.
