Truace
Subscribe
55/100

Moderate evidence

Cite
More
Export CSVExport JSONEmailLinkedInX

Firms that hired more AI workers grew productivity faster after 2018, a study of 1,321 firms finds

Study title: Canaries in the Gold Mine: Early Productivity Gains from Artificial Intelligence Creating Organization Capital

Evidence type
Observational studyPreprint, not yet peer reviewed
Sample
n = 1,321
Sources
1 sourcePeer-reviewed
Published
3 min read
Truvace record
TRV-2026-2510cite or check it

Study title: Canaries in the Gold Mine: Early Productivity Gains from Artificial Intelligence Creating Organization Capital

Researchers at the University of Maryland tracked U.S. public firms from 2010 to 2024, using AI-skilled employment as a measure of AI investment. In the 1,321-firm regression sample, firms that grew their AI-job share more saw faster productivity growth over 2018–2024: 6.2 percentage points more in sales per worker and 7.5 points more in revenue productivity for each standard-deviation increase, roughly one extra point per year.

Read the study abstract

Using a new firm-level measure of AI investment based on AI-skilled employment—spanning machine learning through generative and agentic AI—we show that AI investments are associated with productivity growth in recent years, but not over the previous decade. We trace the productivity gains to the accumulation of organization capital that AI helps create: durable firm-specific knowledge acquired through learning-by-doing that enables more efficient production. We build a novel measure of organization capital based…

The gain

In a study of 1,321 U.S. public firms, a one-standard-deviation rise in the growth of a firm's AI-job share over 2018–2024 went with 6.2 percentage points more sales-per-worker growth and 7.5 points more revenue productivity growth, with no gain over 2010–2018.

In brief

The same firms showed no such link over 2010–2018, so the payoff arrived roughly a decade after AI's commercial rise. The authors attribute it to organization capital, meaning durable firm-specific knowledge built through learning-by-doing, which AI-skilled workers help codify into models, platforms and workflows.

Sales rose while employment did not: a one-standard-deviation increase in AI investment went with 6.7 percentage points more sales growth and a near-zero employment effect. The authors report no sign of rising prices in AI-intensive industries. The findings are associations, though an instrument based on pre-2010 university hiring ties is said to support a causal reading.

Main points

  1. By 2024, U.S. public firms employed more than 300,000 AI-skilled workers, 2.1% of their U.S. workforce; outside public firms the share was about 1%.
  2. Over 2010–2018, AI investment was unrelated to productivity growth; the link turned strongly positive over 2018–2024.
  3. The effect holds for both earlier machine-learning-era AI and generative and agentic AI, not only the recent wave.
  4. AI jobs are almost twice as likely as non-AI jobs to build organization capital (58% vs. 31%); 66% of generative and agentic AI jobs do so versus 57% of earlier machine learning, language and vision jobs.
  5. Growth in organization-capital-building AI jobs predicted productivity growth, while growth in other AI jobs did not.
  6. AI-investing firms expanded intangibles such as SG&A and R&D rather than property and equipment, and shifted toward workers with advanced degrees and senior roles while non-college and junior shares declined.
  7. Employment was unchanged and high-skilled employment rose, which the authors say argues against labor replacement as the main source of gains.
  8. Case studies include Morgan Stanley training an assistant on advisor meeting notes, P&G building a digital twin from plant telemetry and Expedia turning incident histories into agentic workflows.

The full story

AI investment is measured through employment: the authors combined about 456 million Lightcast job postings with Revelio Labs resumes covering roughly two-thirds of the U.S. workforce. They identified 510 AI-related search terms, with a false-positive rate of about 2%, and matched employers to Compustat public firms. The regression sample excludes technology sectors and has 1,321 firms.

The analysis uses long differences, relating changes in productivity to changes in AI-job share, to address the concern that AI-investing firms were more productive to begin with. The effects equal 0.14 of a standard deviation in sales-per-worker growth and 0.19 of a standard deviation in TFP growth. Firms showed no differential productivity trends before 2018, and effects rose steadily as the horizon lengthened after 2018.

Timing tests found AI hiring preceded productivity gains, while earlier productivity growth did not predict later AI investment. Positive effects appeared in manufacturing, wholesale and retail, finance, other non-tech industries and the tech sector. The authors say results are not explained by acquisitions, offshoring, pandemic shocks or non-AI IT investment.

On mechanism, AI-investing firms increased their capitalized-SG&A stock of organization capital over 2018–2024 but not 2010–2018. Gains in productivity and organization capital were larger for firms that started with relatively little of it, as the authors' model predicts.

Sources

1 on file
  1. NBER Working Paperdoi:10.3386/w35684
    Peer-reviewed2026-08-01

Talk it over

Four questions to start onWho gets this benefit, and who was left out of the study or story?

Start with a question

Firms that hired more AI workers grew productivity faster after 2018, a study of 1,321 firms findshttps://truvace.com/item/canaries-in-the-gold-mine-early-productivity-gains-from-artificial-intelligence-

  1. Who gets this benefit, and who was left out of the study or story?

    The gain In a study of 1,321 U.S. public firms, a one-standard-deviation rise in the growth of a firm's AI-job share over 2018–2024 went with 6.2 percentage points more sales-per-worker growth and 7.5 points more revenue productivity growth, with no gain over 2010–2018.

    Answer in the thread

  2. How do we know? What can this kind of evidence show, and what can’t it?

    The evidence This rests on an observational study with a sample of 1,321, from NBER Working Paper. It is a preprint, not yet peer reviewed.

    Answer in the thread

  3. What does it leave open? What else would you want to know before acting on it?

    The record’s own caveat AI investment is proxied by AI-skilled hiring, which does not capture firms relying only on off-the-shelf AI tools, and the sample covers U.S. public firms and excludes tech sectors; results are associations in the baseline analysis.

    Answer in the thread

  4. Have you seen AI used this way where you study, work or live? Did it go the same way?

    Answer in the thread

The questions are the same on every record. The details under them are taken from this record as written.

What readers say

More from Business