Исследования

Speeding up planning decisions with AI

Google DeepMind and UK government work together to reduce planning assessment times

30 October 2026 · Source: Google DeepMind Blog

October 31, 2026 Responsibility & Safety Owen Larter and David Thacker PlanAssist has shown it can cut the time planning officers spend assessing householder applications by up to 58%. PlanAssist - an AI planning tool that handles the heavy lifting of routine applications so officers can focus their expertise where it matters most - is to be made available to an additional 12 local planning authorities across England. In early pilot testing across Barnet, Camden, and Dorset, PlanAssist showed it could bring the average time a planning officer spent on householder assessment tasks down by 58%. Designed to support human judgement, the tool speeds up routine checks while ensuring planning officers retain the final say on every application. Planning bottlenecks have real costs. For the economy, they slow progress toward the UK government's goal of building 1.5 million new homes by 2029. For families, that can mean months of waiting on approval for something as simple as a loft conversion, conservatory, or extension. Built with Gemini, PlanAssist is part of Google DeepMind’s broader partnership with the UK government to explore how AI can help transform planning and other public services. Routine householder applications account for up to 70% of planning officers’ workloads. They're often simple to approve but take up a disproportionate share of their time. Developed by Google DeepMind in collaboration with the UK government, Google Cloud, and Faculty, PlanAssist streamlines this work, giving planning officers automated support with processing applications and drafting assessment recommendations. First, it validates application documents and architectural drawings, checking that the required information has been submitted so the application can proceed. Next, it checks the application against heritage constraints (such as tree preservation orders), and local and national planning policies. It also checks whether any issues or objections have previously been raised, such as by local residents. Finally, it drafts fully cited reports that can be edited by planning officers before they reach a final decision. Since June, the tool has been piloted across Barnet, Camden, and Dorset councils, and has shown encouraging early results. In addition to more than halving the officer assessment time per application, more than 70% of the reports generated by the tool - including things like whether an application complies with local planning rules, or whether the applicant has submitted all required documentation - were approved by officers with no edits required. See how Plan Assist works. PlanAssist is now moving into a 12-month Beta phase using live applications, starting with a subset of councils. Twelve more councils will be onboarded in three cohorts, bringing the total involved to 15. Together, these councils include metropolitan and rural districts across the country, allowing us to test PlanAssist’s capabilities across a diverse range of communities. Subject to successful testing, the government plans for the tool to be made available to all councils in England in 2027, with every assessment reviewed and approved by a qualified planning officer before any decision is made. Planning decisions shape everyday lives, from a family adding a nursery to a homeowner preparing for winter. By handling time-consuming administrative tasks, and ensuring planning officers retain the final say on every decision, PlanAssist shows how AI can help cut backlogs and approve applications faster. Providing access to these new councils is the natural next step, and we look forward to continuing to support the UK government as they seek to build more homes and unlock economic growth across the country.

Read at the source →

More in this topic

ИсследованияSame Facts, Different Remedy: Linguistic Profiling and Arbitrariness in LLM and System One DecisionsarXiv:2611.08995v1 Announce Type: new Abstract: Large language models are increasingly used to resolve consumer complaints, and the way complainants write varies with age, education and socioeconomic status (SES). We audit three remedy-decision systems using sociolinguistic rewrites of US consumer financial complaints that are instructed to preserve every fact: two generative LLMs that return a single label (GPT-4o-mini and GPT-5-mini) and a probabilistic "System One" model that returns a distriarXiv cs.CY (Computers and Society)
ИсследованияAI Exposure and the Changing Responsibilities in Job PostingsarXiv:2611.09870v1 Announce Type: new Abstract: The growing adoption of artificial intelligence (AI) in the workplace has raised questions about how employer demand for different work activities is changing. We examine how the prevalence and content of work activities in online job postings change in association with activity-level measures of AI exposure, derived from Copilot and Claude usage. To enable systematic comparisons, we develop a pipeline that extracts the responsibilities from job adarXiv cs.CY (Computers and Society)
ИсследованияSeeing Is Not Knowing: The Epistemic Limits of LLM-Based Mental Health Inference from Youth Social Media ConversationsarXiv:2611.08565v1 Announce Type: cross Abstract: Adolescents increasingly use social media for private conversations, sharing experiences and seeking support. These interactions generate digital traces, prompting efforts to infer mental health and psychosocial states. As large language models become more capable of reasoning about conversations and interpersonal dynamics, a fundamental question remains nonetheless: what can models actually "know" about an individual's psychological experience farXiv cs.CY (Computers and Society)
ИсследованияEntering a New Domain in the Era of Generative AI: An Autoethnography of Solo Game Development as a Bioinformatics EducatorarXiv:2611.09623v1 Announce Type: cross Abstract: The rapid improvement of generative AI (GenAI) not only increased the efficiency of practice but also empowered individuals with capabilities to explore new domains. However, identity formation in the wake of such empowerment remains understudied. We use an autoethnography to understand identity formation in the context of solo game development as a bioinformatics educator. With the goal of developing a create-your-own-adventure (CYOA) game, we carXiv cs.CY (Computers and Society)