The Coming Collapse of White-Collar Work (buildingbetter.tech)

A provocative warning piece argues that a recent OpenAI “frontier AI” scenario (April 2025) sketches a plausible, near-term pathway where successive generations of autonomous AI agents rapidly automate most knowledge work. The author—an experienced software engineer—summarises a timeline in which models already handle code generation, testing, refactoring and deployment, then evolve into internally trained agents that accelerate their own development. By 2027 the scenario posits a “superhuman” researcher (Agent‑4) running orders of magnitude faster than humans and automating R&D, and by 2029–2030 B2B and open‑source AI offerings make hiring human knowledge workers economically irrational. The piece treats these claims as plausibly imminent rather than purely speculative. The significance is systemic: if white‑collar salaries collapse, the UK faces cascading risks across housing, pensions, welfare, consumer demand, mental health and political stability. The author criticises current UK policy as largely symbolic—taskforces and an AI Safety Institute—contrasting it with binding EU rules and U.S. export controls, and highlights underfunding of alignment, oversight, auditing and AI‑specific reskilling. Technical implications include feedback loops where improved agents speed future model training, compute‑driven capabilities, and commoditisation of expertise as services. The article calls for urgent national action: enforceable transparency and safety standards, large‑scale reskilling or income support, taxing AI rents, and international coordination to avoid a rapid social and economic shock.