---
id: "lib-2026-06-10-ai-economic-indicators-june-2026-update"
type: "library"
title: "AI Economic Indicators: June 2026 Update"
lang: "EN"
org: "Unknown"
source_type: "Report"
date: "2026-06-10"
week: "2026-W24"
url: "https://nosov.ai/library/lib-2026-06-10-ai-economic-indicators-june-2026-update/"
original_url: "https://digitaleconomy.stanford.edu/app/uploads/2026/06/AIEI_RN01_Jun26.pdf"
tier: "A"
tier_name: "Strong evidence"
tier_reason: "Erik Brynjolfsson's Stanford Digital Economy Lab updates its ADP payroll analysis of 25,000 firms with a stated method and adds a 12-indicator takeoff tracker; strong data, released as a short research note rather than a paper."
summary_basis: "full-text"
original_minutes: "20"
tags: ["labor market", "canaries", "automation vs augmentation"]
---

# AI Economic Indicators: June 2026 Update

Tier A (Strong evidence): Erik Brynjolfsson's Stanford Digital Economy Lab updates its ADP payroll analysis of 25,000 firms with a stated method and adds a 12-indicator takeoff tracker; strong data, released as a short research note rather than a paper.

Summary based on the full text.

## In short

Research Note 1 of the Stanford Digital Economy Lab's AI Economic Indicators project (June 2026). It launches three trackers: the Canaries Dashboard (with ADP Research) on employment by age and AI exposure, a Takeoff Tracker of 12 US macro indicators, and an Adoption Monitor collecting surveys. The main finding repeats and extends "Canaries in the Coal Mine": aggregate differences are modest, but early-career workers in exposed jobs keep losing ground.

## Key numbers

- -3.8% vs +2.0%: annual employment growth since November 2022 for workers aged 22-25 in the most versus least AI-exposed occupations
- 1.1% vs 2.0%: annual employment growth for all ages in the most versus least exposed occupations
- 25,000: firms in the balanced five-year ADP sample, employing 4.6 million occupation-coded workers in November 2022
- 7 of 12: aggregate takeoff indicators showing no evidence of AI takeoff as of May 2026; two show strong evidence

## What it reports

- All ages: the most exposed occupations grow 1.1% a year, the least exposed 2.0%, since November 2022.
- Ages 22-25: exposed occupations contract 3.8% a year while the least exposed grow 2.0%; muted similar patterns appear up to age 34. Early-career workers are 7.4% of sample employment.
- Using Anthropic Economic Index usage data, occupations whose AI use leans to automation show declines; augmentation share shows no clear relation.
- Examples: early-career software developers and customer service workers decline; home health aides grow.
- Takeoff Tracker: seven indicators no evidence, three mild, two strong. Capital share keeps rising (strong); TFP growth shows no break (neutral); IP equipment share has recovered to early-2000s levels (mild).
- Adoption Monitor: individual work use generally rising but some recent surveys show a reversal; firm adoption led by the US, with firms expecting more adoption in the next three years.
- Reviews conflicting studies (Yale Budget Lab, Humlum and Vestergaard, Lambert and Schindler and others), stating the question remains debated.

## How they got it

Balanced five-year panel of ADP payroll firms (removing entries and exits), more than 730 occupations ranked by Eloundou et al. exposure scores in quintiles, grouped by age. Takeoff indicators from public US macro series with stated classification rules. Adoption figures compiled from published surveys.

## Limits

- ADP sample is not the full US labour market and differs from QCEW firm composition.
- Exposure measures ignore demand responses to lower prices.
- Correlational; the note does not isolate AI from other causes such as interest rates in this release.
- Short note; details live on the dashboard pages.

## Why it matters for companies

The pattern to plan for is a thinning entry level in exposed functions rather than broad job loss. The automation-versus-augmentation split hints that how a firm deploys AI may shape its junior hiring more than whether it deploys it.

Original: https://digitaleconomy.stanford.edu/app/uploads/2026/06/AIEI_RN01_Jun26.pdf