As the tools and technologies available to planners become more complex and autonomous, their impact on the planning process, profession and practice will inevitably increase. This makes it crucial for RTPI members to have a high-level understanding of how the technologies powering such tools work, allowing us to harness the opportunities they create whilst navigating their many challenges.
This advice note introduces these technologies and some of the tools available for planners and planning systems today. It is for RTPI members to decide how, if, and when to use them.
Contents
- Understanding the technology
Machine learning
Generative AI
Large Language models (LLMs) - Emerging opportunities for planning
Availability of data
Realtime impact modelling
Faster and better consultations
Automated decision making - Emerging challenges for planning
Software as a Service (SAAS)
Data
Copyrighted data
Technology stack
Cyber security - Concluding advice
Existing frameworks
Future considerations
1. Understanding the technology
Most of the technological shifts that have been happening are the result of decades of innovation, initiated by the emergence of computer chips, digitisation, and the global internet.
Digitisation enables the collection and storage of vast amounts of information, in a format which is readable by machines. Powerful computer chips enable all this data to be analysed and processed at incredible speeds. The internet allows for data to be collected and shared globally, and put in the hands of billions of people who are surrounded by computerised devices connected to the web.
The combined impact of advancements in these technologies makes it possible for anyone with access to the internet to have access to artificial intelligence (‘AI’). However, the term ‘artificial intelligence’ itself is a catch-all term for other technological advancements including machine learning, generative AI and large language models.
Machine learning
Machine learning has emerged from the accumulation of vast datasets, combined with advancements in computing power and the creation of innovative algorithms such as ‘neural networks’. Neural networks are computer models inspired by the human brain that specialise in finding complex patterns in data. This evolution has fundamentally transformed the programming of machines, moving away from computer coding for all conceivable scenarios to a model where machines can ‘learn’ and adapt based on their training data. Machine learning is the main technology enabling the autonomous training of algorithms. Algorithms recognize images, text, and speech and streamline the learning process.
Generative AI
Building on the foundation of machine learning, where algorithms can analyse and learn from data, the field of Generative AI has emerged. Leveraging the vast pools of data, the capability to write software now goes beyond learning and recognizing patterns in data to replicate and create new content. This approach is utilised to generate images, audio, text, and videos. AI in this way not only understands but also produces new digital ‘artefacts’.
Large Language models (LLMs)
Large Language Models are a specific type of Generative AI trained on the enormous amount of text available across the internet. LLMs can appear to understand context and generate responses that are coherent and often contextually appropriate. However, it is important to note that these models don't ‘understand’ context in the way humans do. This is because they operate by predicting the next word or phrase that is most likely to follow a given sequence of words, based on patterns learned from the ‘training data’. This process is purely statistical and doesn't involve any comprehension of meaning, intent, or the nuances of human thought (although this is subject to challenge).
At present, there are significant concerns for Intellectual Property rights in the training of LLMs, and in the outputs of Generative AI (see, for example, the UK Government’s 2020 consultation on artificial intelligence and intellectual property).
3. Emerging opportunities for planning
The basic building blocks for AI (digitisation, the web, and computer chips) could improve the efficiency and effectiveness of planners and their impact on the built environment. Some opportunities are presented below.
Availability of data
The process of digitising the significant amounts of planning information being processed on a daily basis will create vast amounts of data. This data will include all existing legislation, policy and guidance, as well as the plethora of data included in the hundreds of thousands of planning applications submitted annually across the UK. We have recently seen significant investment from government in this with the creation of planning.data.gov.uk. This aims to make planning data available to anyone in a standardised and accessible way. Over time it is likely to become a growing source of public sector planning data related to development consents.
Realtime impact modelling
As the amount of data which is available in a format which can be consumed and processed by machines increases, we are likely to see a growing availability and use of real-time impact modelling. Examples range from modelling of housing growth, where application data is linked to live strategic housing land assessments and local housing completions, to modelling on the public health and wellbeing impacts of alternative land use policies for options appraisals and decisions on applications (see, for example, here and here).
Faster and better consultations
Both internet and mobile applications reach significantly larger and harder-to-reach groups of people than physical media. The opportunity to target specific information to specific groups of people and geographies using social media makes it possible to magnify the impact when compared to current methods of consultation.
This may of course result in more public input that need to be processed and analysed, and the emergence of tools that allow such feedback to be automatically grouped and categorised based on sentiments, themes or anything else that might make the analysis of consultation responses more efficient and impactful (see, for example, here).
Critical to many consultation processes is creating visual material for audiences to understand. The use of generative image and video technologies make the creation of this material significantly easier and cheaper, although concerns over the choice or transparency when images are created, and how, needs to be considered. The accuracy of this material is critical, so that development proposals can be seen and experienced as if they were already built.
Automated decision making
The ability for machines to read and understand planning policy might present opportunities for planners to use software to initially screen some types of development applications. In addition, some local planning authorities in England are already engaged in projects to test how policies can best be written and presented in a machine-readable format. This service can help applicants identify where planning permission is needed and apply for simple applications such as Lawful Development Certificates.
4. Emerging challenges for planning
Whilst emerging technologies present a number of opportunities, it is important to understand and consider their potential pitfalls - especially where they might inadvertently create more harm than good. This is also the case for RTPI members in relation to the Code of Professional Conduct.
All RTPI members should remember that technology is always changing, and the pace of this change is increasing. This means that wherever possible you may want to adopt an approach which allows for flexibility and experimentation. Instead of focusing on specific technological solutions, an emphasis on outcomes should ensure that the technology itself will be a means to an end, rather than a means to itself. A number of high-level principles and considerations that could assist are introduced below.
Software as a Service (SAAS)
Most software-based technologies today are ‘cloud-based’. Using a SAAS model, a subscription service negates the need for outright software purchases, reducing any need for maintenance and operational costs. This model can also offer a higher level of flexibility to switch IT of digital planning services as the needs of your organisation evolve. In light of this, opting for shorter-term licences, preferably no longer than a year, could be wise.
Data
Ensuring clarity on data ownership and licensing is of particular importance as it can impact on the ease with which you can switch services. All software interacts with data in various forms—ranging from data supplied to the service provider, to data utilised for training the service, or data generated about the service usage itself. For example, a service that makes data portable by allowing you to download anything you have created or uploaded could make switching services easier.
Copyrighted data
Most AI-based services, including tools such as Chat-GPT or Midjourney, depend on vast amounts of data to train their algorithms. These services often harness data from the internet to enhance their capabilities. However, this practice has led to intricate legal issues, especially when such data is protected by copyright. Consequently, it’s crucial to understand how data you input into these systems are going to be used - specifically, whether your data is employed for analytical purposes, algorithm training, or potentially resold.
Recent legislation for England in the Levelling Up and Regeneration Act 2023 sets out the need for all planning data to be made available under an Open Licence. This will make it harder for software that processes planning data to prevent you from accessing the data you have input in the first place.
Technology stack
In general, IT systems and applications are developed using ‘technology stacks’. These stacks encompass a variety of components, from accessing data through APIs (Application Programming Interfaces) from other providers to particular functions like user management or mapping. The term ‘technology stack’ refers to the collection of technologies and dependencies integrated into the development and operation of a software application. How the stack has been put together can determine the amount of maintenance and how often the IT service goes offline.
Cyber security
Having your services and data on the web (or ‘cloud’) has many advantages, but it also raises security risks. This ranges from others illegally accessing the data, to purposefully making a service unusable. When adopting cloud-based solutions, it is always important to understand what security measures are in place to avoid these things. Measures such as data encryption and two-factor authentication can usually eliminate most threats.
In all cases, it would seem appropriate to seek advice from IT experts within your organisation and potentially also externally from other professionals and consultants with relevant IT experience.
5. Concluding advice
Existing frameworks
We are aware of several frameworks that may be useful in helping planners navigate new technology:
- The Local Digital Declaration has been signed by nearly all English local authorities.
- The gov.uk Service Manual and Technology Code of Practice provides advice on procurement, understanding the technological landscape, and starting with prototypes. It highlights keeping up with technological developments, minimising ownership costs, using standard government components, ensuring data control, and incorporating open standards for easier future adjustments.
- The Scottish Government has launched the Digital Skills Portal; this online tool aims to enhance the digital confidence of public sector planners in Scotland and has features of relevance to planning professionals working in other parts of the UK. The portal includes a diagnostic questionnaire for users to self-assess their level of digital confidence. This identifies gaps in knowledge and experience across the three themes of policy-making, enforcement and data monitoring.
- The Planning Inspectorate has published advice on using AI in casework evidence.
- LexisNexis, in partnership with Dr Su Chadwick of Pinsent Masons, has produced a comprehensive practice note on digital planning in England which covers key legislation. This is behind a paywall, but your organisation may have access.
Future considerations
In the world of post-AI technology, planning is facing a transformative shift. For RTPI members, gaining a solid grasp of these underlying technologies is becoming essential. However, as with most technologies, the rapid evolution and power of AI will mean that the impact it has on planning will also be evolving. This is most evident in LLMs and generative design.
Given the scale of the shift this represents, the Institute is looking at conducting further research to develop a deeper understanding of the impacts that AI is having, or may have, on how RTPI members plan, consult, and make decisions.
All RTPI members should continue to remind themselves of the requirements of the Code of Professional Conduct.
