Harry Quartermain is RTPI South West Chair and Head of Research at LandTech
Our plan-led planning system is under significant strain. Around 80% of planning authorities in England are currently operating under the presumption in favour of sustainable development, either because their local plan is out of date or because of historic under-delivery on housing. Planning reforms introduced over the last 18 months are now compounding this pressure: concurrent local government reorganisation, the re-introduction of strategic planning through Spatial Development Frameworks, and a new 30-month local plan review process are all happening simultaneously.
At the same time, the planning profession and the public alike are rapidly up-skilling on generative AI tools. The question planners need to ask is not whether this is happening, but what it means for how the system functions.
The 30-month process and the consultation burden
The government's expectation is that local plans will be prepared and adopted within 30 months. Under this new framework, authorities face multiple rounds of mandatory consultation: a scoping consultation before the clock even starts, a further consultation within the first year on proposed plan content and the evidence base, and a third consultation on the near-final draft before Gateway 3. Each of these rounds will generate a substantial volume of representations from land promoters, planning agents, and members of the public.
That volume is about to get significantly larger, and the signal-to-noise ratio significantly worse.
AI-generated objections: problem or participation tool?
A growing number of members of the public are now using generative AI to draft planning objections and representations. Commercial services exist specifically for this purpose, generating objection letters that cite policies and case law for a fee. The problem is well-documented: large language models are prone to confabulation, asserting the relevance of cases or policies that do not exist, or mischaracterising those that do. Every representation, however, must be considered under statutory obligations and in accordance with the Gunning Principles. The burden this places on already stretched planning officers is real.
But it would be too simple to characterise AI-assisted engagement as straightforwardly harmful. There is a legitimate argument that these tools lower the barrier to participation for people who would otherwise find the planning system impenetrable: jargon-heavy, procedurally complex, and apparently designed for specialists. If AI gives a resident in a local community the confidence to engage for the first time, that is not nothing. The challenge is distinguishing genuine place-based insight from templated bulk submissions that carry no material weight and consume officer time disproportionate to their value.
The planning profession needs to develop a clearer position on this distinction, and the Planning Practice Guidance needs to catch up with the reality planners are already dealing with.
AI talking to AI: the next frontier
The public are using AI to object. Planners are beginning to use AI tools to assist with processing, summarising, and responding to representations. We now have a situation in some areas where AI-generated objections are being triaged and responded to by AI-assisted officer processes. What does meaningful public participation look like at that point?
My view is that this dynamic is neither inherently good nor bad, but it does require the profession to be intentional rather than reactive. The risk is a kind of arms race in which the substance of local plan making, the genuine negotiation between competing interests about how places should grow and change, gets buried under automated processes on both sides. The opportunity, if AI is used well by planning authorities, is that it frees officer capacity to focus on exactly that substance: the evidence, the trade-offs, the genuinely contested decisions that require professional judgement.
The key question is who controls the AI, to what end, and with what transparency. An authority that uses AI to process representations efficiently and dedicates recovered time to meaningful community engagement is using technology in the public interest.
The case for structured engagement
Given all of this, there is a strong case for local authorities to invest in short-form, targeted, evidence-led engagement as a supplement to statutory consultation. Structured surveys, designed with clear demographic and geographic intent, produce quantified data on local attitudes in a format that is significantly easier to analyse, report, and defend as part of a sound evidence base.
This matters for two reasons. First, it helps insulate the consultation record from the distortive effect of high-volume, low-signal submissions, whether AI-generated or otherwise. Second, and more positively, it gives planning authorities a robust, representative dataset that genuinely reflects the views of their communities, not just those who are motivated and technically equipped to engage through traditional routes.
The planning system is being asked to do more, faster, with less. AI is not going away, and neither is the public's use of it. The profession needs to adapt now: investing in smarter engagement, clarifying what counts as a valid representation, and using technology on its own terms rather than simply absorbing its consequences.
