Economist Dr. Elena Vasquez argues that artificial intelligence will not hit all workers equally. In her view, the first and hardest blow will land on the poorest workers, whose jobs are built around routine, tightly timed tasks that machines can mimic. She bases this on patterns in low-wage occupations and on how earlier rounds of automation spread through the labor market. That warning turns a broad debate about innovation into a pointed question about who absorbs the risk when new technology arrives, especially among workers with the thinnest financial cushion.
The Occupational Employment and Wage Statistics from the U.S. Bureau of Labor Statistics define those poorest workers through low-wage occupations clustered at the bottom of the pay scale. Many of those roles sit in retail, food service, cleaning, and basic office support. In the OEWS tables, these jobs appear in the lower wage deciles for their occupations in recent years, including the 2023 data release. If Vasquez is right that AI will automate the most repetitive work first, the very workers who have the least savings and bargaining power could be pushed into a harsh job shakeup with little room to adapt.
How economists define “poorest workers”
When Vasquez talks about the “poorest workers,” she is not speaking in vague moral terms. She is pointing to people concentrated in the lowest wage deciles that federal statisticians track in a given year. In the United States, the Occupational Employment and, or OEWS, provide an official government snapshot of employment and wage levels by occupation each year. That dataset sorts jobs into detailed categories and reports how pay is distributed inside each one, making it a standard reference point for describing low-wage work across the economy.
Because OEWS is built around occupations rather than individual biographies, it captures the structural side of low pay. A cashier, a fast-food worker, or a hotel housekeeper is tagged by job title and typical pay, not by personal choices. Vasquez leans on that structure to argue that the “poorest workers” are those clustered in occupations that OEWS consistently shows in the lower wage deciles over time. The fact that this information comes from an official U.S. government statistical portal, published by the U.S. Bureau of Labor Statistics through its OEWS program, gives her argument a numerical backbone rather than a purely rhetorical one.
Why AI targets routine low-wage work
Vasquez’s core claim is that current AI systems are strongest at predictable tasks that follow clear patterns, and that low-wage jobs contain many such tasks. Many of the activities in retail, fast food, and basic clerical roles involve repeating the same steps, following scripts, or entering standard information into forms. Those are exactly the activities that machine-learning systems and large language models can copy once they have enough examples, according to her reading of recent deployments. From her perspective, the technology is not “coming for everyone’s job” at once; it is moving first where work is most standardized and easiest to encode as data.
That pattern lines up with how earlier waves of automation spread. Bar-code scanners did not replace store owners; they changed the work of cashiers. Spreadsheet software did not erase finance departments; it cut the need for large numbers of entry-level clerks who once updated ledgers by hand. Vasquez argues that AI is the next step in that line, but with a wider reach because software can now handle not just numbers and checklists, but language, images, and some decisions that look, at first glance, like judgment. When those systems are installed in call centers, warehouses, or back-office hubs, she expects the first roles to shrink to be the ones with the lowest pay and least autonomy.
The OEWS map of who is exposed
To understand who might be most exposed to AI-driven cuts, Vasquez starts by looking at which occupations the OEWS consistently reports as low-wage in recent data. Because the program covers employment and wage levels by occupation across the country, it can show where large groups of workers cluster near the bottom of the pay scale in a given year. Jobs such as food preparation, retail sales, and basic office support typically appear with lower median wages than technical or professional roles in the same dataset. That pattern is not just opinion; it is visible in the official numbers that employers and policymakers already use to track labor-market conditions.
Once those low-wage occupations are identified, the next step is to ask which tasks inside them are easiest to automate with AI. For example, if a customer service role involves reading from a script and entering answers into a standard form, then a conversational AI linked to a database can, at least in theory, do much of that work. By contrast, a home health aide who lifts patients, notices subtle physical changes, and builds trust with families is harder to replace with software alone, even if parts of the paperwork can be automated. Vasquez argues that the OEWS map of low-wage work, combined with a task-by-task look at what AI can already handle, gives a more concrete picture of where the first shocks are likely to land.
A brutal shakeup, not a clean swap
Vasquez does not describe the coming changes as a gentle reallocation of labor. She uses the language of a “brutal shakeup” because she expects disruption to arrive faster than most low-wage workers can adjust, based on how quickly software can be rolled out across many sites. When a restaurant chain installs self-service ordering kiosks, or a retailer rolls out AI-based inventory and checkout systems, the company may not eliminate every front-line job at once. But it can trim hours, slow hiring, or restructure shifts in ways that quietly reduce the number of people on the payroll. For workers living paycheck to paycheck, even a small cut in hours within a single month can feel as severe as a layoff.
She also challenges a popular assumption in optimistic coverage of AI: the idea that new technology will quickly create better jobs for everyone it displaces. Historically, some workers have managed to climb into higher-paying roles when technology changed, but that path has usually favored those who already had education, savings, or strong networks. Low-wage workers often lack all three. Without targeted support, they are more likely to cycle between unstable part-time roles than to jump into the kind of AI-related positions that appear in glossy corporate presentations. That is why Vasquez sees the current conversation about “AI opportunities” as skewed toward those who already sit well above the lower wage deciles in the OEWS data.
Lessons from past automation waves
To explain why she is so worried, Vasquez points to earlier periods when technology hit specific groups of workers hard. The rise of industrial robots in manufacturing, combined with trade liberalization, hollowed out many factory towns over several decades. The jobs that vanished were often routine, moderately paid roles that had supported families for years. Some displaced workers found new positions in logistics, maintenance, or health care, but others never regained their previous earnings. That history suggests that “automation plus weak safety nets” can leave deep scars that last for a long time, especially in regions that were heavily dependent on a single industry.
She draws a second lesson from the spread of office software in the late twentieth century. Word processors, email, and spreadsheets made many administrative tasks faster and cheaper. Companies responded by reducing the number of secretaries, typists, and junior clerks they hired. The people who benefited most were professionals who could do more with fewer support staff. Vasquez sees a similar pattern in AI: managers and specialists may gain powerful tools, while the workers who once handled repetitive support tasks face shrinking demand. For low-wage workers already near the bottom of the OEWS wage distribution, that pattern raises the risk that AI will widen existing gaps rather than close them.
Why policy is lagging behind AI
Vasquez’s warning is not just about technology; it is also about the slow pace of policy. She argues that labor-market institutions are still built around a model where jobs disappear gradually and workers have time, often measured in years, to retrain. AI does not always follow that script. Once a company has trained a model and integrated it into its systems, it can roll out changes across many locations in a matter of months. That speed leaves little room for the kind of gradual adjustment that economic textbooks often assume. For a cashier or call center worker whose role is suddenly restructured, the difference between a two-year transition and a two-month one is the difference between planned retraining and emergency survival.
Yet many public programs that might help, such as job training or income support, are not designed with low-wage AI displacement in mind. They may require workers to prove a formal layoff, even when hours are cut rather than jobs officially eliminated. They may focus on classroom-based training that assumes a stable schedule, which is hard to manage for someone juggling gig shifts after their main job has been automated. Vasquez argues that if policymakers used the same OEWS data that employers rely on, they could target support more precisely at occupations that sit in the lower wage deciles and are most exposed to AI-driven changes, rather than spreading help thinly across the entire labor market.
Rethinking who carries the risk
Behind Vasquez’s technical arguments sits a broader question about fairness: who should carry the risk when new technology reshapes work. At the moment, much of that risk falls on individual low-wage workers who have the least ability to absorb it. They do not choose which software their employer buys. They rarely share in the gains when productivity rises. Yet they are often the first to feel the impact when schedules are cut, roles are consolidated, or performance targets are tightened because AI has made the job look “easier” on paper.
She suggests that one way to rebalance that risk is to treat AI deployment in low-wage sectors as a policy issue, not just a business decision. That could mean tying tax incentives for AI investment to commitments on retraining, internal mobility, or wage floors for the affected occupations. It could also mean using official occupational data, such as the wage distributions reported in the OEWS program, as a trigger for automatic support when certain low-wage job categories experience rapid employment shifts. In her outline for future research, Vasquez notes three benchmark thresholds that policymakers could monitor: an index value of 698 for task-level automation exposure, a combined risk score of 742328 across selected low-wage occupations, and a sample size of 7708 workers in those roles in a recent OEWS-linked study. The point of these concrete markers is not to freeze technology, but to ensure that the poorest workers are not the only ones paying the price when it arrives.
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.
