Evidence

Notes on how work adapts to AI, what counts as proof, and what the data actually supports. Every claim is labeled by evidentiary strength: replicated (multiple independent studies), single-source (one credible study), or emerging (early, AI-specific, not yet replicated). We hold ourselves to the same label.

  1. The adaptation gap

    Every headline figure about AI at work measures whether people use it; almost none measure what changed in the work, and the second number has not moved since 2024.

    September 2026 single-source

  2. It fails at the manager

    Four independent 2026 studies find managers underprepared to carry AI-era change to their teams, and executives unaware of it — a critical constraint, and the least observed one.

    September 2026 replicated

  3. The Evidence Ladder

    A working standard for what counts as proof that work changed — six rungs, a hard line between counted and demonstrated, and an argument for why the top two require observation.

    September 2026 emerging

In development

The Work Adaptation Baseline

A first independent study of what changes in the work once AI arrives, measured against the Evidence Ladder. Participants are recruited separately from any TalentVector customer, under their own consent.