The Adoption Reality Check: What Five Years of Pew Data Shows About American Workers and AI

Executive Summary

• Five years of Pew Research Center surveys — the most credible independent longitudinal dataset on American public AI opinion — show adoption growing but sentiment deteriorating. The two trends are running simultaneously.
• 21% of U.S. workers used AI in their job as of September 2025, up from 16% a year earlier. The remaining 65% use it minimally or not at all. (Pew ATP, n=5,010 workers, September 2025)
• 50% of U.S. adults are more concerned than excited about AI in daily life — a 13-point increase since 2021. Only 10% are more excited than concerned. (Pew ATP, June 2025)
• The expert-public gap is the widest it has been in five years. 56% of AI experts anticipate positive impacts over 20 years; only 17% of the general public agree. That 39-point gap is the backdrop against which every enterprise AI rollout occurs.
• Organizations using Pew data as a planning input — rather than vendor surveys that oversample AI optimists — will make more realistic adoption timeline and change-management budget decisions.

The Adoption Picture: Who Is Actually Using AI at Work

The September 2025 Pew survey of 5,010 employed U.S. workers — a probability-based nationally representative sample — produces the most reliable denominator in the workforce AI adoption literature.

21% of U.S. workers use AI in their job. That’s up from 16% a year earlier — a 5-point gain in 12 months. But the structure of that gain matters: the share who say “some” of their work uses AI rose from 14% to 19%. The share who say “all or most” of their work uses AI stayed flat at 2%. AI is entering the workflow at the edges, not restructuring jobs at the core.

65% of workers use AI minimally or not at all. This isn’t a laggard tail — it’s the center of mass. The 21% who use AI are the outlier cohort, not the baseline.

Education is the most significant predictor of current adoption. Workers with a bachelor’s degree or higher: 28% use AI (up from 20%). Workers with some college or less: 16% (up from 13%). The gradient is real and growing.

Age follows education. ChatGPT workplace use by age cohort: 38% of workers aged 18–29, 30% of those 30–49, 18% of those 50+. Senior employees — often the decision-makers and most experienced contributors — are the lowest-adoption group.

36% of non-users believe their work could incorporate AI (up from 31%). 46% believe their roles cannot be automated. The non-user population is not uniformly resistant — it is split between the curious-but-untrained and the genuinely insulated. These populations require different interventions.

The Sentiment Picture: Concern Is Outpacing Adoption

The adoption curve is rising. The sentiment curve is not rising with it.

50% of U.S. adults say the increased use of AI in daily life makes them more concerned than excited. Only 10% are more excited than concerned. 38% sit at rough parity. The concern-dominant majority has grown 13 percentage points since 2021 — from 37% to 50%.

This matters for enterprise AI leaders because your workforce reflects the general public until your change-management program overrides the baseline.

On AI’s impact on jobs specifically: only 23% of Americans are optimistic. By contrast, 44% are optimistic about AI’s impact on medical care. The jobs number has not moved meaningfully across Pew’s five-year longitudinal series. Job-impact anxiety is sticky, not cyclical.

The expert-public gap is the widest signal in the dataset. 56% of AI experts anticipate positive impacts over 20 years. 17% of the general public share that view. That 39-point gap is not narrowing as adoption rises. More exposure is not automatically producing more confidence.

Near half of both experts and the general public feel they have little or no control over AI in their lives — and majorities in both groups want more oversight. Even advocates for AI acknowledge they don’t feel in control of it. That’s the organizational psychology your change-management program is working against.

Daily AI interaction is growing faster than workplace adoption. 31% of Americans interact with AI several times daily, up from 22% in early 2024. This is mostly consumer AI (Siri, Alexa, Google, ChatGPT for personal use). The consumer familiarity doesn’t automatically translate to workplace adoption — but it does mean the workforce is not AI-naive.

Age, Education, and the Demographic Reality

Under-50 workers are the primary AI user cohort. Approximately half of adults under 50 interact with AI daily or more. The 50+ cohort — which includes most senior managers, many directors, and a significant share of VPs — is the lowest-interaction group.

The generational distribution creates a structural deployment tension. In most organizations, AI tools are being used most by individual contributors and least by the managers evaluating their output. The BCG finding that managers role-modeling AI use is the #1 predictor of team adoption (BCG AI at Work, n=10,635, 2025) maps directly onto Pew’s age-stratified adoption curve: the people whose modeling behavior matters most are the people least likely to be using AI.

The awareness plateau is real but doesn’t equal readiness. 47% of Americans have heard “a lot” about AI — a 21-point increase since 2022. Broad awareness is no longer the constraint. Structured, role-specific training that converts awareness into workflow habit remains the gap.

Key Data Points

Source credibility: HIGH. Pew Research Center is a nonpartisan, nonprofit fact tank. American Trends Panel is probability-based and nationally representative. No vendor funding, no commercial interest in AI adoption direction. This is one of the most methodologically credible sources in the corpus.

What This Means for Your Organization

The Pew data provides the baseline your change-management program needs to calibrate against. Most enterprise AI planning uses adoption surveys commissioned by vendors or consulting firms with commercial interest in optimism. Those surveys find 70-90% adoption intent. Pew’s population-representative panel finds 21% actual use and a majority of the workforce more concerned than excited.

The practical implication is sequencing. If your workforce reflects national demographics — and at 200–2,000 employees, it is far more demographically representative than a Fortune 500 engineering team — you are likely planning against a baseline where 65% of your employees have minimal AI engagement and 50% are net-concerned. That’s not a laggard workforce. That’s the American workforce in September 2025.

The expert-public gap (56% vs. 17% optimistic on 20-year impact) matters for your internal communications. Executives who consume AI industry media have been absorbing the expert-optimistic narrative for years. Many of their employees have not. The perception gap BCG and Wharton documented between executives and middle managers has a population-level foundation in Pew’s data — it is not a management failure, it is a base-rate problem.

Three specific interventions follow from this data. First, age-stratified onboarding: the 50+ cohort — most likely your senior managers — needs role-specific entry points, not general AI literacy sessions designed for early adopters. Second, the 36% of non-users who see potential in their role is the highest-leverage cohort for rapid adoption expansion. These employees are not resistant; they lack structure. Third, the 46% who believe their role cannot be automated should not be pushed — they are likely correct, and confronting them with job-displacement framing will confirm their worst concerns about AI’s intent.

If the Pew framing surfaces questions specific to how your organization should sequence change management and training investment, that conversation is worth having — brandon@brandonsneider.com.

Sources

1. Pew Research Center, “Key Findings About How Americans View Artificial Intelligence” (March 12, 2026). Synthesis of five Pew ATP survey waves, 2021–2026. URL: https://www.pewresearch.org/short-reads/2026/03/12/key-findings-about-how-americans-view-artificial-intelligence/. Credibility: HIGH — nonpartisan, no commercial interest, longitudinal, probability-based.

2. Pew Research Center, “About 1 in 5 U.S. Workers Now Use AI in Their Job” (October 14, 2025). American Trends Panel (ATP), n=5,010 employed U.S. workers, fieldwork September 2–8, 2025. URL: https://www.pewresearch.org/short-reads/2025/10/14/about-1-in-5-us-workers-now-use-ai-in-their-job/. Credibility: HIGH — same panel, primary data, specific to workforce adoption.

3. Pew Research Center American Trends Panel (ATP) — probability-based online panel, nationally representative. Multiple waves referenced: June 2025 (public sentiment, n=~8,750), August 2024 (domain-specific optimism), February–March 2025 (ChatGPT workplace use by age), 2024 (expert survey, methodology not specified in synthesis).

Brandon Sneider | brandon@brandonsneider.com April 2026