Skip to content
AI: Jobs, Power & Money
17MAR

One in three US workers now uses LLMs

3 min read
13:50UTC

LLM adoption among American workers rose from 30.1% to 38.3% in twelve months — faster than smartphones at the same penetration stage — but whether that adoption is replacing jobs or reshaping them remains genuinely contested.

PoliticsAssessed
Key takeaway

At the current adoption growth rate, LLM saturation among US knowledge workers is likely within two to three years.

LLM adoption among US workers rose from 30.1% in December 2024 to 38.3% by December 2025 — an 8.2 percentage-point increase in twelve months, according to Hartley et al. 1. For comparison, US smartphone penetration took roughly four years to traverse the equivalent band in the early 2010s.

The adoption figure alone does not settle the displacement debate. Oxford Economics found no evidence of AI replacing workers at scale . The NBER's Humlum and Vestergaard found task restructuring "without net changes in hours or earnings" 3. The "AI washing" research suggests many companies are citing AI for cuts driven by conventional cost pressure.

Block's experience tells a different story — its CFO cited a measurable productivity gain, then the company cut 40% of its workforce. Both accounts — augmentation in some firms, genuine replacement in others — can coexist. That is precisely what the aggregate data obscures.

The skills gap makes the stakes plain. The workers being hired are not the workers being displaced. If the Hartley et al. adoption curve continues, more than half of US workers will use LLMs by late 2027. Whether that correlates with accelerating displacement or continued augmentation depends less on the technology's capability than on corporate incentive structures — and equity markets are currently rewarding headcount reduction over workforce expansion.

Deep Analysis

In plain English

Hartley et al.'s data tracks how many US workers use large language models like ChatGPT or Claude in their jobs. The share rose from roughly three in ten in late 2024 to nearly four in ten by end of 2025 — an 8-percentage-point gain that is fast by historical technology adoption standards. Two gaps matter beyond the headline figure. First, the survey measures whether someone has used an LLM, not how often or how deeply — a worker who tried it once counts the same as one using it six hours daily. Second, adoption is almost certainly concentrated among higher-income, more-educated workers, meaning productivity gains visible at the individual level are not evenly distributed across the workforce.

Deep Analysis
Synthesis

The 38.3% adoption figure, combined with the NBER finding of no net changes in hours or earnings, points to an adoption-without-displacement pattern consistent with productivity diffusion lags. The more diagnostically significant dynamic is that adoption is increasing the variance of worker outcomes rather than shifting the mean: AI-intensive adopters are gaining measurable advantages while the majority show no aggregate change — a pattern that widens inequality before any net job displacement occurs.

Escalation

The 8.2 percentage-point annual gain is likely to decelerate as adoption reaches the later-majority segment — workers with less digital fluency or in roles with less discretionary task structure. Employer mandates, as at Accenture, could steepen the curve before it flattens. Adoption in the 55–60% range would represent saturation of the addressable knowledge-worker population.

What could happen next?
  • Meaning

    Adoption above 38% means LLMs have crossed from early-adopter to early-majority status — historically the inflection point at which a technology becomes a baseline job requirement.

    Immediate · Assessed
  • Risk

    Concentrated adoption among higher-income workers could widen wage inequality even without net job losses, as AI-augmented workers command growing salary premiums over non-adopters.

    Short term · Suggested
  • Consequence

    Employers mandating AI adoption will increasingly treat LLM proficiency as a baseline hiring criterion, disadvantaging workers without access to paid AI tools or digital training.

    Short term · Suggested
  • Opportunity

    The 61.7% of US workers not yet using LLMs represents a latent productivity reserve that targeted employer training programmes could unlock at relatively low marginal cost.

    Medium term · Suggested
First Reported In

Update #1 · Meta cuts 20% while Big Tech spends $650bn

Fortune· 17 Mar 2026
Read original
Different Perspectives
Oxford Economics
Oxford Economics
Concluded AI's role in recent layoffs is 'overstated,' finding companies are not replacing workers with AI at scale. Identified slowing growth, weak demand, and cost pressure as the actual drivers.
Ambrish Shah, Systematix Group
Ambrish Shah, Systematix Group
Warned AI coding tools will erode Indian IT firms' labour-arbitrage growth model by reducing enterprise dependency on large vendor teams.
South Korean government
South Korean government
Enacted the world's second comprehensive AI law, choosing an innovation-first framework over prescriptive employment protections — a deliberate contrast to the EU's regulatory approach.
Corporate executives executing AI-driven cuts
Corporate executives executing AI-driven cuts
Frame workforce reductions as existential necessity. Crypto.com CEO Kris Marszalek and Block CEO Jack Dorsey both described AI adoption as a survival imperative, with equity markets reinforcing the message through immediate share-price gains.
Chinese government (Wang Xiaoping)
Chinese government (Wang Xiaoping)
Positions AI as a job-creation engine to absorb 12.7 million annual graduates and offset 300 million retirements, directly contradicting domestic economist Cai Fang's warning that AI job destruction precedes creation.
Klarna and companies reversing AI cuts
Klarna and companies reversing AI cuts
Klarna's public reversal — rehiring the human agents it replaced with AI after customer satisfaction collapsed — validates Gartner's prediction that half of AI-driven service cuts will be undone by 2027.