AI could automate most UK remote work by 2030, government report warns
Key Points
- A UK government report says AI could automate most tasks performed by remote workers by 2030
- The report was published by the Government Office for Science on 15 June 2026
- It sets out five scenarios across three AI capability trajectories
- Around 70% of UK jobs are exposed to AI, the highest among advanced economies
- The scenarios are designed to help policymakers stress-test policy, not to predict outcomes
Artificial Intelligence could automate most tasks performed by a remote human worker by 2030, according to a new UK government report.
The Government Office for Science (GO-Science) published AI Scenarios 2030 on Monday (15 June), produced with the AI Security Institute (AISI) and the Department for Science, Innovation and Technology (DSIT).
The report sets out five scenarios for how AI could develop by the end of the decade, grouped into three capability trajectories in which the rate of progress has either slowed, continued, or rapidly accelerated.
In three of the five scenarios, AI reaches a point where it can automate most tasks a remote worker could perform.
In the most advanced scenario, “Take-Off”, leading systems outperform expert humans at virtually all cognitive tasks from 2029 onwards, with significant advantages in domains such as software engineering, AI research and development, and cybersecurity.
How fast AI capabilities are moving
The report tracks capability gains using data from METR (Model Evaluation & Threat Research), which measures the length of software engineering tasks that frontier models can complete autonomously at around a 50% success rate.
That figure rose from approximately four minutes in March 2024 to 12 hours as of February 2026, with early signs it could extend to weeks-long tasks.
The report notes the benchmark is task-specific, measured at partial reliability, and does not imply equivalent autonomy across real-world domains.
GO-Science states that AI systems crossed critical thresholds in 2025, citing Anthropic’s finding that systems were able to substantially increase the risk of severe misuse.
It adds that capital expenditure by the largest technology firms more than doubled between 2023 and 2025, while ChatGPT’s user base expanded from roughly 200 million weekly users in mid-2024 to around 800 million by late 2025.
The job market impact
The report found that around 60% of jobs in advanced economies are exposed to AI, with the UK’s service-based economy among the most exposed at around 70%.
Exposure measures the degree to which an occupation’s tasks align with AI capabilities, not the extent to which roles will be automated.
In the higher-displacement scenarios, “Transformation Economy” and “Take-Off”, AI is primarily substitutive for human labour rather than complementary.
The report states that in these futures only half of those replaced find new employment, with many forced to accept wage reductions and less secure work.
Entry-level recruitment in many sectors collapses, removing entry points for young people, while demand for experienced experts holds.
In the “Take-Off” scenario, AI pushes unemployment above recession levels alongside rapid economic growth, producing what the report describes as a “K-shaped” economy.
In the more moderate “Augmented Growth” scenario, social, legal, and practical considerations, including compute constraints, keep humans “in the loop” for most tasks, and many new roles emerge in AI integration and oversight.
Market concentration and geopolitics
The report expects the frontier AI market to remain highly concentrated toward 2030, with a few large technology companies dominating development.
It notes that NVIDIA occupies 92% of GPU market share and that the largest training runs are likely to cost over a billion dollars by 2027.
GO-Science also points to a shifting geopolitical landscape, noting that Chinese firm DeepSeek released a system in January 2025 trained at a fraction of the cost of Western competitors, with China since becoming the leader in open-weight systems.
The report notes that this has effectively ended America’s uncontested AI leadership.
GO-Science stresses that the scenarios are not predictions but tools for policymakers to stress-test policy, and that the real future is likely to combine elements of several scenarios. The report does not estimate the likelihood of each.
