economics1 min read

Are we worrying about the wrong workers?

By Pedro Serôdio

Employment in the UK occupations most exposed to AI has grown since ChatGPT launched, while the least exposed have declined. The fall is concentrated in construction and other physical work, alongside sharp declines in planning approvals and housing starts.

Are we worrying about the wrong workers?

Employment in the UK occupations most exposed to generative AI has grown by 5% to 10% since ChatGPT launched, depending on the exposure measure, while employment in the least exposed occupations has fallen by 2% to 4%. The decline is in occupations that were supposed to be safe from automation: bricklayers, scaffolders, machine operatives, warehousing and delivery. Low-exposure service work such as care and hospitality grew over the same period, so the losses are specific to work tied to building and physical infrastructure.

This piece attributes that decline to the cost of carrying out physical activity in Britain. Detailed planning approvals in England are down around 37% since 2019, brick deliveries across Great Britain around 30% since 2018, and London housing starts fell from 21,700 in 2022-23 to 4,200 two years later. Research carried out with Britain Remade finds UK infrastructure costs 65% above OECD comparators. How many of these jobs exist depends on how much gets built, connected and maintained, a quantity set by planning committees, grid queues and payroll costs.

§ Related

Related Articles

0012026Even superhuman AI may not replace jobswriting
0022026Is AI the new breadwinner?writing
0032026AI and the UK labour market: the evidence so farwriting
Rev. 08.2026email@pedroserodio.comLondon, United Kingdom
Contact