UK launches AI to speed up planning for 1.5M new homes

The UK government, in collaboration with Google, has announced a new artificial intelligence system named "Extract" aimed at significantly speeding up the planning permission process for new homes and infrastructure projects.[1][2] Unveiled during London Tech Week, the initiative leverages Google's Gemini AI model to digitise and analyse planning documents, a traditionally cumbersome and paper-reliant task.[1][3] This move is a core component of the government's strategy to meet its ambitious target of building 1.5 million new homes over the next Parliament and to modernise the nation's planning system.[1][4] The "Extract" system is designed to bring England's often outdated planning procedures into the 21st century by tackling the extensive time currently spent by planning officers manually reviewing vast quantities of information.[1]

The new AI tool, "Extract," developed by the government's AI Incubator team (i.AI) with support from Google, is set to revolutionise how local councils process planning applications.[1][3] Currently, around 350,000 planning applications are submitted annually in England, many of which involve hundreds of pages of paper documents, including decades-old handwritten notes and maps.[1][5] Planning officers are estimated to spend around 250,000 hours each year manually checking these documents.[1][5] The "Extract" system, utilising Google DeepMind's Gemini model, can scan and convert these complex documents, including blurry maps and handwritten text, into usable digital data in a fraction of the time it currently takes.[1][3][4] Test trials conducted in councils such as Hillingdon, Nuneaton & Bedworth, and Exeter have demonstrated significant efficiency gains.[1][6] In these trials, "Extract" digitised planning records, including maps, in approximately three minutes each, a task that typically takes one to two hours manually.[1][5] This suggests that the AI tool could process around 100 planning records per day, drastically reducing delays that have long plagued the system.[1][5] The system uses Gemini's multimodal reasoning capabilities to interpret text, images, and handwritten annotations, and then employs tools like OpenCV and Ordnance Survey for polygon extraction and georeferencing, converting historical maps into modern, geolocated versions.[3] The government aims to make "Extract" available to all councils in England by Spring 2026, with plans to expand its capabilities to handle all types of planning documents.[1][5]

This AI initiative directly addresses the UK's ongoing housing shortage and the government's "Plan for Change" milestone to construct 1.5 million new homes.[1][4] The current planning system is widely acknowledged as a significant bottleneck, with slow approvals hindering development and exacerbating the housing crisis.[7][8] By automating time-consuming aspects of the planning process, such as compliance checks and the digitisation of historical records, the government anticipates that planning officers will be freed up to focus on more complex decision-making, thereby accelerating the approval of new homes and critical infrastructure projects like wind farms and data centres.[1][7][8][9] The government estimates that the digitisation of the planning system, including initiatives like "Extract," could lead to annual time and cost savings of £527 million for the public sector.[1][9] The Prime Minister, Keir Starmer, stated that the outdated planning system has held the country back and that "Extract" represents a bold step towards a 21st-century planning system.[1][6] Similarly, Deputy Prime Minister and Housing Secretary Angela Rayner emphasised that leveraging cutting-edge technology like "Extract" can help fix the broken planning system, cut delays, save money, and reduce burdens on councils.[1][9]

The collaboration between the UK government and a global tech giant like Google on such a critical national issue has significant implications for both the AI industry and the broader adoption of technology in the public sector.[3][5] It signals a growing trend of governments turning to AI to solve complex societal challenges and improve public service delivery.[3][8] The choice of Google's Gemini model underscores the advanced capabilities of modern AI in understanding diverse data types, from text to technical drawings.[3][5] Demis Hassabis, Co-Founder and CEO of Google DeepMind, expressed excitement about the UK government choosing Gemini to speed up the planning process.[1][3] Google will also support the national scaling of "Extract" through reliable and secure hosting on Google Cloud, along with providing AI and engineering advice.[3] This project is seen as a stepping stone for exploring similar AI applications in other public sector areas.[3][8] However, the increasing reliance on AI in public services also brings considerations around data privacy, the accuracy and potential biases of AI models, and the need for human oversight.[10][11] Some councils have already raised concerns about the potential for "AI-driven bulking" in applications and the impact on public participation if AI is not used transparently.[10] The Alan Turing Institute has also highlighted that while AI can detect common errors and digitise records, human oversight by planning professionals remains crucial.[12][11] A Ministry of Housing, Communities and Local Government study found AI response accuracy ranged from 43% to 76%, emphasising this need.[11]

In conclusion, the launch of the AI-powered "Extract" system, developed by the UK government and Google, represents a significant attempt to modernise the country's planning permission process and accelerate house building to meet pressing national targets.[1][2] By leveraging Google Gemini's advanced capabilities to digitise and analyse complex planning documents rapidly, the initiative promises to reduce bureaucratic delays, free up planning officers' time, and contribute to substantial public sector savings.[1][3][9] This project not only tackles the immediate challenge of housing supply but also serves as a prominent example of public-private collaboration in deploying AI for critical infrastructure and public service reform, potentially paving the way for wider AI adoption across government while also highlighting the ongoing need for careful consideration of ethical and practical implementation challenges.[3][8][10]