What AI is really doing to people at work.
Measured with evidence, published in the open.

Why we publish

Evidence about AI and people should be public, not buried in a slide deck.

Research agenda

Six open questions. One for each domain.

Our research follows the framework. Each domain starts with a question we expect the evidence to answer, one assessment at a time.

IDignity

What does constant tracking do to people?

AI makes it easy to track, time and score work. We want to know where that helps people, and where it wears them down.

IIAgency

Do people still get a say when AI decides?

As more decisions are made or shaped by AI, we look at who keeps the final word, and what happens to judgement over time.

IIITransparency

Are people told before AI changes their job?

Disclosure is easy to promise and hard to do well. We compare what companies say they tell people with what people say they were told.

IVResponsibility

Who answers when AI gets it wrong?

We look for a named owner, a real way to challenge an outcome, and a record of putting things right when it matters.

VFairness

Who keeps the productivity gains?

AI can raise output without raising pay or prospects. We track how the gains and the costs are shared between groups.

VIAdvancement

Is AI building skills or hollowing them out?

We look for evidence that people are learning, moving into new roles and progressing, not just keeping up.

What we’ll publish as the evidence grows

Three formats, from early notes to an annual index. Each one waits until the evidence can carry it.

Notes

I · From the first assessments

Field notes

What we are learning about the method itself: what worked, what didn’t and what we changed.

Briefs

II · Next

Benchmark briefs

Short, sector-by-sector comparisons, once enough companies in a sector have been assessed.

Index

III · Later

The Humanistic AI Index

An annual, independent view of how AI is landing on people across sectors, built from aggregated evidence.

How we research

Rigorous and anonymous. Open about method.

  1. IBuilt from assessments

    Our research draws on the evidence gathered in assessments, aggregated across companies.

  2. IIAnonymous by default

    No company is named without its consent, and survey results are only reported for groups large enough to protect identities.

  3. IIIMethod first

    We publish how we reached a finding alongside the finding itself, so anyone can check it.

  4. IVHonest about limits

    Early samples are small. We say so, and we never claim more than the evidence supports.

The research road

Nothing is published yet. This is the order it will come in.

  1. IFounding assessmentsThe first body of independent evidenceNow
  2. IIField notesEarly lessons on the methodNext
  3. IIIBenchmark briefsSector comparisons as the dataset growsNext
  4. IVHumanistic AI IndexAn annual, cross-sector viewLater

Get the first report

Leave your email and we’ll send our first research as soon as it’s published.