Across the UK, councils are increasingly adopting AI planning validation checks in an effort to improve efficiency. On paper, the idea is attractive. Automation promises quicker registration, lower administrative burdens and a more consistent approach. In reality, however, many applicants and agents are finding that these systems are slowing the planning process down rather than speeding it up. Rather than reducing bureaucracy, validation is becoming more onerous.
The importance of Judgement at validation
The problem is that validation is not simply a mechanical exercise. It requires judgement. Experienced officers have traditionally been able to recognise the difference between a genuinely invalid application and one which, while perhaps imperfect in presentation, contains enough information for the council to assess it properly. They could apply proportion and common sense. After all the purpose of planning application documents is to provide enough information for the council and members of the public to understand what is proposed and the potential issues with an application and no more. It is not an exercise in extracting more information (whether relevant or not) for the sake of it.
AI-led systems, or rigid automated processes built around them, often cannot do that. Instead, they tend to identify technical discrepancies, formatting issues, divergences from guidance or minor omissions as reasons to reject or delay registration. Documents may be present but labelled differently. Plans may provide sufficient information but not in precisely the expected format. Requirements for documents may be made where there is no need. A human validator may once have accepted such submissions as providing sufficient information and moved matters forward. An automated system is more likely to trigger delay.
The result is more invalidation notices, more correspondence, more frustration and more wasted professional time. Far from reducing pressure on council departments, this can actually increase workloads by creating unnecessary rounds of checking and resubmission. It also means more costs.
Technology in Planning
Technology has a role to play in planning administration, but it should support officers rather than replace judgement. Used properly, AI can help identify obvious gaps and improve consistency. Used too rigidly, it risks creating a validation process that is more bureaucratic, less efficient and less sensible than the one it was meant to improve.
At Planning Direct we believe the government’s new emphasis on using technology to improve the planning process is welcome but actual real-world impacts need to be understood. We are not AI experts (though this author has managed a few IT projects). We understand that judgement in grey areas is crucial. AI does not exercise judgement to the degree experienced planners do. But judgement in the planning process needs to be exercised from the validation stage, not just at decision time. The introduction of AI into validation is exacerbating an existing issue – the overzealous application of validation requirements where the resulting information is irrelevant or unnecessary and certainly of no use in decision-making. The ‘computer says no’. This results in more cost for all with no added benefit.
What do we need to do?
If the government wants to get the best from new tech in government processes it needs to ask the same questions asked ahead of any IT project. What outcome do I want? At present the introduction of AI into the planning process is increasing costs for all and impacting timeliness. That surely is not the outcome intended?
The Government may also be making an age old mistake in IT rollout. New tech is no good without new training. Council planning departments have for long said that they do not have enough staff or enough time to train them. WIthout the right training the undeniable administrative benefits of AI will not be realised; indeed at present it seems AI has made things worse.