Wave 1 · fielding · Q4 2026 embargo target

State of AI Adoption in Knowledge Work

Skill premiums, task replacement vs augmentation, and proof-of-work by job title. This page is methodology plus a live instrument. Headline percentages will not appear until the wave closes.

12-minute survey Read the method

Integrity rule. No illustrative statistics. If you see a chart with percentages on this URL before we announce close of fielding, treat it as a bug. Pay and hours saved are self-reported perception, not payroll match. We will footnote that on every published table.

Targets (plan, not results)

1,200 completes. At least 120 per function cell. Mix of US, UK, Canada, and India. Seniority split: IC / senior IC / manager. Soft-cap a cell at 200 so marketers cannot drown PMs.

Product

180

Marketing

180

Analytics

180

Finance

150

Ops

150

CS / CX

180

HR / L&D

180

Bars are target n, not collected n.

Citeable tables we will publish

The questionnaire is designed backward from these outputs. If a question does not feed a cell, we cut it.

ID	Finding	Breaks	Likely linker
T1	Median self-reported AI skill premium vs same-title peers	Title × seniority × region	HR / comp blogs
T2	% whose last performance review mentioned AI work	Function × manager vs IC	People-leader newsletters
T3	Hours saved per week (median, p25, p75) among weekly users	L1–L5 × role	Productivity press
T4	Task map: replace vs augment vs no-AI	Role × task	L&D / bootcamps
T5	Tool concentration: primary model + weekly stack size	Role × company size	Vendor-neutral roundups
T6	% who can show a before/after metric vs “I use ChatGPT”	Competency level	Career coaches
T7	% of job-switchers asked for an AI artifact	Role × company size	Recruiting media
T8	Blockers ranked: policy, skill, time, trust, access	Industry × size	Enablement blogs

Methodology

Opt-in web sample, not a probability panel. We will say that in paragraph one of the report.
Weight by role, seniority, and region. Publish effective n and design effect.
No causal claims on pay. T1 is a relative band, not matched payroll.
Competency is scored from behaviors (schema, eval, agents, governance), not from “I am advanced.” Same L1–L5 as the PM guide.
Release: HTML report + CSV of aggregate tables + embeddable charts with required source line. Microdata stays off-site.
Screener: employed in knowledge work, 20+ hours/week.

How the sample is built

Primary: this form, plus the homepage quiz (“help build the benchmark”). Secondary: partner newsletters with a unique UTM and a function-cut they can republish. Incentive is the early PDF, not a raffle. Speed and duplicate checks on.

Wave 1 instrument

About 12 minutes. Email is optional and used only to send the report. See privacy.

About you Job family Seniority Region Company size

Use and outcomes Days per week you use generative AI for work Primary tool Hours saved last week (self-report) Did your last performance review mention AI work? Relative pay vs same-title peers (self) Can you show a before/after metric for one workflow? In your last interview cycle, were you asked for an AI artifact? Top blocker

Behaviors (L1–L5 scoring)

How true is each, at work, in the last 90 days? 1 = not at all, 5 = routinely.

I use a saved schema or checklist on a repeating task (not a fresh chat) I score AI output against a written rubric or test set A human must approve before AI output ships to customers or leadership I route more than one AI job with handoffs (not a single prompt)

Contact (optional) Email for the report One workflow you would show a manager (optional)

Embargo plan

Week	Asset	Ask
−3	This methodology URL indexed	Nothing to embargo yet
−1	T1–T4 charts + CSV + quote cards	Link to interactive tables, not the homepage
0	Report hub + 7 role cuts	Partners who helped field get an exclusive function chart
+1	Skill-gap calculator using T4 weights	Embed with required attribution

PM transition SOP · How to read salary claims · 8-question quiz