The evidence desk of ALECAM

The record on AI, work, and the people between them.

Evidence reviews and annual reports on adoption, readiness, and workforce risk — measured, sourced, and stated plainly. Every figure carries its instrument.

03entries in the record

33+data sources in Part I alone

2024–26evidence window

No.Review / key findingsKey figure

01Evidence Review
Part I

AI Use Across Generations, Professions and Disciplines

What thirty-three data sources establish about who uses artificial intelligence, who does not, and what the difference actually measures.

49% of U.S. adults used AI chatbots in February 2026, up from 33% in mid-2024.Pew / Feb 2026

Adults 50+ using generative AI rose from 9% in 2023 to 30% in 2025 — a tripling in two years.AARP / 2023–25

34% of law-firm professionals use AI tools their firm has not authorized.Thomson Reuters

September 2026 · Version 1.0 · 33 sourcesRead the review →

02Evidence Review
Part II

The International Comparison

How U.S. patterns of AI use compare with Europe, Asia and the emerging economies — and which findings survive the crossing.

Europe’s youngest-to-oldest age gradient is 9.1:1; the comparable U.S. “ever use” gradient is 2.9:1.Eurostat / Pew

64% of European non-users say they did not feel the need to use generative AI.Eurostat / 2025

U.S. population-level adoption is 28.3%, ranking 24th on the Stanford measure.Stanford AI Index / 2026

September 2026 · Version 1.0 · companion to Part IRead the review →

03Annual Report
First Edition

The State of AI Workforce Readiness 2027

Why adoption stopped being the metric that matters — and what replaces it. The Human Above the Loop™ Report.

24%strongly agree their organization prepared them to use AI effectively

Jan 1, 2026generative AI exclusions enter standard commercial liability policies

5 tiersof verified proficiency define the workforce-readiness standard

Published Q4 2026 · Novelle & Associates · Human Above the Loop™Read the report →

Method

How we read the numbers.

Adoption figures for materially similar questions can range from roughly 18% to 89% depending on who is asked and how. The Record holds that spread honestly — every figure carries its instrument, population, and reference period.

Vendor and self-selected surveys are treated as unreliable and flagged, not averaged in.
Generational differences are read as associations until controls for occupation, education, income, firm size, and job level are applied.
Where sources diverge materially, the divergence is shown rather than reconciled.
Population adoption, employee use, enterprise deployment, and platform telemetry remain separate units of analysis.