Overview
Generative-AI exposure identifies where tasks may be technically susceptible, not who will adopt. Across European workplaces, conversion depends on capability, task fit, employee influence, digital intensity, and training; a UK attitudes study adds perceived social risk. Adoption is associated with task change, but causation and lasting transformation remain unresolved, making equitable work a question of access, agency, legitimacy, and accountability.
Learning outcomes
- Exposure identifies technical potential, but adoption depends on how workplaces convert that potential into usable practice.
- Education, abstract tasks, organisational influence, digital intensity, and training are associated with stronger conversion, not proven universal causes.
- The European gender gap is concentrated in the most exposed occupations; UK evidence also makes perceived social risk part of the adoption decision.
- Adoption is associated with task restructuring, but a shift-share analysis found no detectable average AI effect on reported task displacement or creation.
- Equitable adoption calls for access, learning, worker influence, accountability, and safeguards—not pressure to maximise use.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
What is the difference between AI exposure and AI adoption at work?
AI exposure is the technical potential for tools to affect job tasks, while AI adoption is whether a worker reports using AI at work. Exposure does not establish access, permission, intensity, safe use, or job transformation.
Why do similarly AI-exposed workers adopt at different rates?
Similarly exposed workers adopt at different rates because capability, task fit, workplace influence, digital infrastructure, training, and perceived social risk shape whether potential becomes usable practice. The studies identify associations, not a single causal mechanism.
Why might a capable worker choose not to use generative AI?
A capable worker may decline generative AI because the tool lacks a suitable workflow, permission, time, accountability, or acceptable safeguards. Concern about privacy, accuracy, labour, environmental, or mental-health effects can be a considered judgement rather than a skills deficit.
Does higher workplace AI adoption mean jobs are being transformed?
Higher reported adoption does not by itself prove job transformation. Adoption and task restructuring are strongly associated, but the early shift-share analysis found no detectable average AI effect on reported task displacement or creation.
What does equitable AI adoption at work require?
Equitable AI adoption requires access, time to learn, output-evaluation skills, meaningful worker influence, clear accountability, and safeguards for privacy and other social risks. The evidence supports this diagnosis, not a universal intervention or maximal-use target.