Despite recent calls for a pause on AI development, AI is beginning to transform the way we work across many industries; so how might it help our planning system in the near future? A system which has become notoriously slow, document heavy and undoubtedly affected by political fragility.
ChatGPT described the planning system to me as “slower than a sloth on tranquilizers” and provided a summary of how AI could be used to improve the system, although failed to understand the practical issues faced by local planning authorities and applicants. ChatGPT’s response however highlights just how quickly AI is able to generate accurate and informed content, even if not always perfect.
My analysis is that there are 8 areas where we should expect to challenge ourselves with AI tools:
- Planning application validation – the current system allows applicants to submit invalid applications, and requires a manual review by officers which is incredibly time-consuming and leaves applicants waiting for weeks to find out their application is in fact invalid (an enormous 50% of householder applications are invalid). AI could detect errors instantly to prevent invalid submissions, and entirely take over the validation role; this would simultaneously free up officer time and provide a more user-friendly experience for applicants.
- Data analysis – AI can process large amounts of data quickly and accurately. Planning is a document heavy industry, and a significant amount of officer and consultee time is spent trawling through reams of documents. AI could be used to summarise application documents, identify missing information or issues, and perhaps assess compliance with policy (depending on how policies and application materials are set up). AI could also be used to generate a skeleton committee report, particularly as several elements of the report are factual (planning history of the site, statutory and neighbour consultations, relevant policy etc.). ChatGPT pointed out that AI could also incorporate some witty one-liners to inject some humour into the system…
- Modelling – AI will improve the way we assess the impact of proposed developments on the environment. AI can, if properly trained, provide instant feedback on how a scheme will affect everything from air quality, to traffic and wildlife habitats, allowing planners to review accurate data on the impacts of the development without necessarily needing the page count of explanatory text that creeps into the process. How that analytical process is kept transparent so that it can be made accountable will be a key challenge, though.
- Decision Making – decision making in planning is based on legislation, planning policy, and judgement. Whilst AI may be able to easily summarise and assess applications against policy, judgement is a fundamental part of planning and (presumably) one that should not be taken over by AI. Issues in planning are often not a straightforward yes or no answer and so the human element will still be required. However, AI could be used to make decisions or recommendations on discrete areas such as certificates of lawful use/ development or condition discharge applications, which are more factual exercises, and even extend to determining minor prior approval applications.
- Legal Agreements – the rumour is that AI is coming for us lawyers. Whilst we already make use of AI, advances with AI will reduce the time lawyers spend on administrative tasks and allow agreements to be created, considered and completed more quickly. There will still undoubtedly be negotiations between lawyers on commercial points, but AI may be able to take the lead on bespoke drafting. It will still ultimately require training and decent precedents to learn from, though. AI will also be able to change the way that research is carried out; however, only once judgments are written by AI will AI be the best tool to interpret them.
- Local Plan examination – the Planning Inspectorate confirmed in August 2022 that it had awarded a contract to Oxford Global Projects UK to explore a range of uses for AI in the examination of local plans, particularly around the standardisation of Local Plans. The average Local Plan examination requires 10-15 days of hearings, whilst the average time to adopt a Local Plan is 19 months; if AI is trained to assess a Local Plan, it has the potential to significantly reduce the examination process and allow Local Plans to be adopted sooner.
- Resourcing issues – it is no secret that local planning authorities are significantly under-resourced. Planners are stretched to their limit, and yet need to spend large amounts of their time undertaking administrative tasks. Resourcing issues result in applications getting stuck in the planning system, and delayed decisions, therefore delaying delivery of necessary housing. The use of AI will help unlock staff resourcing issues and hopefully make planners’ lives a lot easier! The investment needed to achieve that will be huge however and is unlikely to be self-funding for several years.
- Limitations – much like humans, AI is sometimes wrong! Testing ChatGPT with generating a basic planning agreement and it included a paragraph stating the Council may serve a breach of condition notice in the event of a breach of the section 106 agreement, which clearly isn’t an applicable enforcement mechanism. The knowledge cut-off date is September 2021 which means it is also entirely unaware of planning policy changes post September 2021. Obviously, these issues may be mitigated or removed over time, but for now AI is not at the stage where it can be relied on to be correct or up to date. In addition, AI is only as good as the data it is trained on. If the data is biased, the AI will adopt that bias and make biased decisions. Planning decisions are heavily influenced by politics; AI would therefore not be able to identify where a decision was made for political reasons.
To conclude, in the words of ChatGPT, “we can expect to see AI playing an increasingly important role in shaping the future of the UK planning system.” There are certainly limitations for the use of AI within the planning system. However, AI is undoubtedly going to take over time-consuming administrative tasks and improve the planning system for the better; the only question is, how and when will the resources needed to unlock it be available?