No real-time observatory of AI-driven labour displacement
Nobody in the UK can currently see displacement as it happens. DSIT's AI and Future of Work Unit (first assessment Jan 2026) concedes causal links between AI adoption and employment change are unknown; ONS surveys were never designed for task-level, high-frequency measurement; a 2021 occupational reclassification truncates usable panels to ~20 quarters (British Progress). One-off studies (KCL's firm-level finding of -5.8% junior roles; IFOW graduate work) are snapshots. No linked dataset combines HMRC PAYE real-time information, job postings and firm-level AI adoption.
Every transition instrument (retraining, income bridges, place-based plans) depends on knowing which occupations and places are being hit, quarterly not retrospectively. The US-centred debate is being imported unverified; UK exposure (~70% of workers, per DSIT) is higher than America's.
A UK AI Labour Market Observatory with statutory access to HMRC RTI payroll data, ONS microdata and commercial postings feeds, publishing quarterly occupation- and place-level early-warning indicators; housed in ONS or as a What Works-style centre, feeding the AI and Future of Work Unit and Skills England.
// Build together: Counterparty: ONS/HMRC via existing SRS/Datalab access plus commercial postings feeds; statutory RTI access is the end-state.
Every retraining, income and place-based response depends on quarterly displacement data no one produces; the linkage is buildable now within existing secure-research access as graduate hiring visibly cracks.
One gap, several dossiers: entries folded into this one (2)
The research pass surfaced this gap independently in more than one domain. Those entries are merged here so the map counts it once: one observatory on linked HMRC real-time and vacancy data, seen from the labour-market, skills-system and policy lenses.
№ 43 · No real-time, granular skills and AI-exposure data infrastructure (Education & AI)
Skills England's 2026 annual report concedes labour-market data 'lacks sufficient detail at local and occupational levels'. No official ONS series tracks AI's effect on vacancies or task composition; the sharpest signals on collapsing graduate/entry-level hiring come from private job boards and the Institute of Student Employers, with contested causality. DfE's Unit for Future Skills was absorbed into Skills England without an open real-time data service emerging. Policymakers, providers and career services cannot see where AI is displacing entry-level tasks until years later.
Its fill: An open National Skills Data Service: linked HMRC RTI, LEO outcomes and vacancy-scrape data published as local occupational dashboards, with a maintained AI-exposure early-warning index; buildable by ONS/Skills England or an independent institute with API access mandated.
№ 165 · No AI transition strategy: displacement monitoring and worker support are unowned (Policy (lens))
The UK has no AI statute: government has declined to back the Lords AI (Regulation) private member's bill (last updated April 2026), preferring regulator-led rules. More concretely, there is no labour-market transition plan. DSIT's own assessment of AI capabilities and labour-market impact concedes evidence gaps on which occupations face displacement and how fast workers can transition; British Progress and LSE analyses find no displacement at scale yet, but the British Chambers of Commerce warned in April 2026 that 'Britain's workforce is not ready'. Skills England is new and oriented to current shortages; DWP has no AI-displacement monitoring function. No institution owns the question of what happens if displacement arrives quickly.
Its fill: An AI Workforce Transition Observatory (DWP/DSIT/ONS) publishing real-time occupational exposure and vacancy data; a funded retraining entitlement targeted at high-exposure occupations; and pre-committed policy triggers (support packages that activate if displacement indicators cross thresholds). Committee interest: Work and Pensions Committee, SIT Committee.