Local plan consultations can generate thousands of responses from members of the public, the development industry and statutory consultees. For local planning authority (LPA) teams, that means weeks of reading, categorising and summarising before the real work of analysis can begin.  

With the government committed to faster local plan preparation, we're exploring whether artificial intelligence (AI) can help reduce the time spent on some of the administrative tasks without compromising the quality of the consultation analysis while continuing to support greater citizen engagement.  

Where this started 

Through support from our PropTech Innovation Fund, we funded Greater Cambridge Shared Planning (GCSP) – the shared planning service for Cambridge City and South Cambridgeshire District Councils – to explore how AI could transform the consultation summarisation process. GCSP partnered with the University of Liverpool to develop a tool. 

Working with the planning team, the University built a bespoke large language model (LLM) trained on 15 years of Cambridge’s consultation data and key planning terminology. 

In a trial last year, the tool, called PlanAI, successfully summarised 3 planning consultations, on supplementary guidance documents, in 16 minutes – a task which took 18.5 hours to do manually. As well as the significant time difference, the quality of the summaries was also compared, with the AI-generated reports providing clear breakdowns of the themes, patterns and insights from the responses.  

The tool works by taking each consultation submission and recognising which comments or parts of comments relate to different policy areas – such as affordable housing or climate change – or chapters such as housing and environment. It then reviews all the representations together and can produce summaries by chapter, policy or site allocation, drawing out valuable insights. It can also provide a spatial representation of the comments, helping teams to identify under-represented areas that may need to be targeted for future engagement.   

By handling the time-consuming task of grouping and summarising consultation comments, the tool enables officers to spend more time reading and analysing feedback, and engaging directly with their communities and stakeholders.  

Taking it further by scaling the initial pilot 

The results achieved in Greater Cambridge were really promising but we wanted to know if they could be replicated in other authorities, and achieve the same speed, accuracy and quality.   

To find out, the Digital Planning programme provided a further £45,000 to GCSP to stress-test the tool with 5 more local planning authorities – varying in size, location, consultation platform and stage of plan-making. 

 The 5 local planning authorities involved are:  

Royal Borough of Greenwich  Hull City Council  Milton Keynes City Council   Liverpool City Region Combined Authority  Tameside Metropolitan Borough Council 

Terry de Sousa, Planning Policy and Strategy Team Leader, Greater Cambridge Shared Planning, said: “We are delighted to share the work we have developed with the University of Liverpool with 5 other planning authorities. After 18 months of rigorous development, evaluation and testing, we now have a shared ambition to stress test the PlanAI in a range of contexts across the country, ensuring it is robust and fit for purpose at each stage of plan-making.”

Progress so far 

In the last few weeks, the 5 local authorities have begun testing the tool on a range of local plan consultations – some recently closed, others more historic. They've provided baseline data on how long previous consultations took to summarise manually, and planning officers are reviewing the AI-generated summaries to ensure they properly capture the consultation responses. This is an important step before relying on the tool more widely. 

Marco Chan, Planning Policy Officer, Royal Borough of Greenwich, said: “Utilising AI technology in planning consultation is an exciting opportunity as it could boost efficiency and ensure an objective point of view in analysing main points in a multitude of feedback. For a local plan public consultation this could mean saving weeks in summarising representations and afford officers more time to refine policies and work on other documents.”  

Professor Alex Lord, University of Liverpool, said: “Working with the pilot authorities has really illustrated the wide variety of approaches local authorities take when conducting and reporting on public consultations. The positive news is that the LLM-based approach we have developed has worked well across all the pilot authorities. As we progress the technology we’re now in a much better place to cover the full extent of varying local authority practices.” 

Building in transparency and trust 

A key part of the University of Liverpool's research has been exploring the ethical, environmental and transparency considerations that come with using AI in this way. During their early work, when they were developing the tool, they held focus groups with members of the public and planning stakeholders to understand people's questions and concerns early on. 

The findings have guided how GCSP communicates about the tool and builds confidence in its use. This has also led to the creation of an 'instant summarisation tool', which GCSP has been using at the front end of the process for their local plan consultation – summarising comments as people submit them, and giving respondents the option to amend the summary if they wish before it is submitted to the councils. 

What we hope to achieve 

This pilot has been about stress-testing the technology to see if it can deliver the same efficiencies and accuracies in different settings. At the end of March, the authorities involved will produce a blueprint report to support other LPAs looking to adopt similar approaches.  

Our ambition is for local planning authorities to be able to explore the use of AI to support faster local plan preparation. 

If you want to learn more about the pilot and hear about the progress so far from the teams involved, join our online show and tell at 1pm on 26 February. Register to receive an invitation.  

For more information about the Digital Planning programme, follow us on LinkedIn to stay connected, and subscribe to our newsletter for the latest updates.