.png)
What makes new tools and approaches for linear infrastructure planning
trustworthy?
Overview of different types of advanced technologies and approaches, applications in
infrastructure space, legal and regulatory requirements, and key issues for trust
The Chair summarised the key issues in the draft briefing paper. She noted that there are a range of technologies used in planning, including established ones like GIS systems and advanced ones like AI. Not all AI is the same. As set out in the diagram, there are two basic types of AI: rules-based (programmed by people, cannot learn) and data-based. Advanced algorithms are increasingly being combined with different AI types, particularly machine learning (e.g., deep learning). Concerns around AI models going out of control are more relevant to frontier models, and not immediate concern in linear infrastructure planning.
There are many current and potential applications of AI in linear infrastructure planning:
Current uses: AI is already used for data collection, spatial planning, reasoning, project scheduling, costing, scenario development, options development, engagement platforms, visualisations, and training.
Ongoing development and potential applications: include decision support tools, social impact assessment modelling, addressing resource constraints, and transforming supply chains.
The Chair then outlined some of the potential risks of AI and associated trust issues in Linear
Infrastructure Planning:
Data-related Risks
Concerns about biases in data sets, exclusion of certain data types, and poor-quality data usage.
Issues with rules and assumptions in models, categorisation of terms, and clear understanding of model impacts.
AI impacts related to safety, security, and personal data/privacy risks.
AI regulation may be needed, but it is emergent.
Sectors like energy and water are already heavily regulated. Challenges in coherence between technical and spatial planning regulation may need to be addressed before
Trust needs accountability, assurance, and transparency; trust-building requires good relationships between stakeholders.
Need to consider trust at different stages of technology lifecycles and usage: upstream (data), model itself, downstream (outputs).
AI assurance/accreditation schemes are still developing – it may be better to rely on existing AI standards e.g. for the public sector.
Cultural and normative aspects play a role in addressing AI-related challenges and
The following points were made in discussion:
The rapid pace of technological change makes it challenging for non-tech-savvy
The pervasive nature of technology and AI could lead to a "battle of the tools". This
Examples of good practices were mentioned by panel members:
The Longfield Solar Farm project - a physical model was used to demonstrate how the solar farm would fit into the landscape. It showed details like tree height, rights of way network, and the substation's location. The community engagement process involved virtual modelling and creating a physical model, which allowed the community to provide feedback. This highlights the importance of explaining technology in a way that people can relate to.
The use of digital tools as a game for community engagement in Germany and
How can tech developers best demonstrate that their technologies and processes are
trustworthy?
The Panel explored this question by using a real example of advanced algorithms/AI in
infrastructure planning; Continuum Industries’ Optioneer routing tool.
Continuum’s CEO, Grzegorz Marecki outlined how in the early stages of a project, significant
decisions are made with limited data, which can have a lasting impact on the project's later
stages. As the project progresses, more information becomes available, but the ability to
influence decisions decreases. Continuum Industries aims to address this by providing early
access to data insights, enabling better decisions and minimising environmental impacts.
Grzegorz acknowledged the delicate balance between openness, transparency and proprietary concerns, describing it as a "chicken and egg" situation.
Continuum Industries is a private, for-profit start-up funded by angel investors and venture capitalists. There are challenges in balancing openness with the need to protect IP. Additionally, the company relies on geospatial data, often subject to restrictive licenses, which limits its ability to share data. Grzegorz also emphasised the importance of collaborating with utilities/developers, and Continuum’s customers, who dictate the level of openness and confidentiality in each project. In Continuum’s view, there is a lot of discussion around trust and AI but there are many layers to that. It’s not just about algorithms, it’s also about how data is presented and processed what goes into a model and who controls that.
Next, Andy Malekos, Head of AI at Continuum, made the distinction between white box and
black box algorithms. Black box algorithms (e.g. deep neural networks) are often complex and opaque. The inner workings are not easily understandable or interpretable by humans.
Understanding why a black box algorithm makes a specific decision can be challenging due to its complexity. The output is accepted based on the model's performance rather than an explicit understanding of how it arrived at that decision.
White box algorithms are transparent and have a clear structure (e.g. linear regression where the relationship between inputs and outputs is straightforward and can be expressed in a simple equation). The logic and decision-making process are explicit. White box algorithms are interpretable, making it easier for humans to grasp how the algorithm arrives at a particular decision. White box algorithms are suitable for tasks where simplicity and interpretability are crucial, like decision trees used in classification problems.
Optioneer's algorithm is an evolutionary population-based algorithm that fits in the white box category. The algorithm explores and optimises different solutions for infrastructure
assets, simultaneously considering factors such as cost, technical feasibility, and environmental impact. Andy highlighted the benefits of this approach, emphasising that it remains simpler and more explainable compared to black-box neural network algorithms.
Andy identified three ways Optioneer helps with explainability:
Freedom to configure everything. This involves allowing users to input their assumptions and preferences before the algorithm runs, enabling them to see a clear link between in-puts and out-puts and giving them the final say over outputs.
Powerful and intuitive visualisations, such as heat maps, provide a high-level overview of different solutions, making the results more intuitive and understandable and exposing trade-offs. These can help build a narrative around the results coming out of the model.
Giving users the final say and control ensures that the algorithm's output is a starting
Break-out groups then discussed what tech developers can do to demonstrate that their tools and processes are trustworthy in terms of in-puts, models and outputs. The following themes emerged in the discussion:
Establishing principles and transparency
Need for a common and visible set of principles guiding data origin, acquisition and
model assumptions. Given fast pace of change a principles-based approach was
thought more appropriate than having fixed rules.
Advocated for openness to facilitate the use of local data that reflects citizens'
challenges and opportunities.
Importance of demonstrating the linkage of outputs to specific community inputs,
creating an audit trail back to the model's origins.
Stressed that the development process should extend beyond outputs, integrating into a broader stakeholder engagement process to build trust over time.
Improving data input and output quality and integrity
Importance of transparent data selection for the model and coordinated data sharing.
Need for standardising data across models and balancing open-source options with
Explored the role of government regulation and oversight in ensuring effective
Addressed output quality, emphasising sense-checking, visually comprehensible results and the use of live data for interrogation.
Flagged importance of providing support appropriate to users’ needs so they can
It was noted that the counterfactual was a situation where you can ‘loose the truth’ as all sides are overwhelmed by information.
Addressing issues of trust, legitimacy and accessibility
Concerns about the transparency of tools when used by developers.
Assurance processes need to be ongoing.
Discussed potential solutions, such as funding community access to digital planning
Recognised the importance of making tools accessible to non-developers.
Highlighted the need to expose underlying values and foster productive disagreement to prevent hidden values from fuelling conspiracy theories.
Acknowledged the financial and broader implications of these considerations.
What steps can other stakeholders take to help ensure that the advanced tools and processes used in infrastructure planning are trustworthy?
Eric Brown, Executive Adviser at Energy Systems Catapult, shared his experiences from the
energy sector in terms of achieving trustworthy tools and processes from the perspective of
project developers/utility companies, and policymakers/regulators. His overarching point was that none of these groups can solve trustworthiness alone. Ongoing collaboration is essential for sustainable progress in the ongoing energy system transformation.
Eric emphasised that it is crucial to understand that tools and processes are aids, not decision-makers. People still need to make the final decisions, emphasising the need for clarity in stakeholder engagement processes and the need to place these in a particular context. The Electricity Networks Commissioner Nick Winser’s work on reducing the time to develop electricity transmission lines is an example. Key recommendations include a strategic spatial energy plan and agreed design principles, contributing to faster decision-making. There is also a focus on corridor planning and route planning, all involving a human decision-maker.
Project developers and utilities play a crucial role in ensuring trustworthiness. They must avoid isolating problems and consider the broader system transformation. Using established plans, principles, standards, and best practices is key. Additionally, ensuring that the process for selecting tools is robust is essential for fostering confidence. Users should be clear on their tool's role, mindful of liabilities and responsible in their use. Good data, proper training and governance are essential, with continuous improvement aligned with stakeholder needs. From the perspective of policymakers and regulators, principles, and practices, accountability and governance remain vital. Clear guidelines are needed for tool use and audits, reviews and compliance measures to ensure responsible application. Collaboration with stakeholders and adapting policies and regulations to align with tool use is crucial. Lastly to protect citizens' interests, addressing ethics and data protection, encouraging data sharing, supporting training and transparent two-way communication are essential. Stakeholders should feel confident that these measures align with the energy system's purpose.
Break-out groups then discussed what steps stakeholders other than tech developers can take to help ensure that the advanced tools and processes used in linear infrastructure planning are trustworthy. The following themes emerged in the discussion:
Policy and regulatory ambiguity, standardisation and collaboration
● Concerns about ambiguity in the roles of regulators (e.g. differences between sectors)
and the need for clear explanations of each regulator's specific role.
● The importance of clear policy narratives and priorities and effective communication
between all stakeholders. There are lots of policy steers in technical planning but fewer
in spatial planning.
● Time constraints limiting further exploration of certain issues in-depth – this is a
common challenge across sectors and reinforces the need for standardisation and
collaboration.
● The pace and scale of change will lead to a concertinaing of local and national planning
processes – this should encourage a more fundamental review of processes.
● Questions about the number and sources of tools, the possibility of a government-
mandated/standardised tool funded by entities like Ofgem.
● In energy, it should be noted that the FSO will be a public corporation with associated
accountability, assurance and transparency requirements.
Integration of technical and spatial considerations
● Technical and spatial planning are not currently integrated - addressing this challenge is
considered vital for improving the system's effectiveness.
● Feasibility of adopting a true whole systems approach, considering the incorporation of
micro and macro elements, indicating a concern for comprehensive integration.
Data gathering and mapping challenges
● Inadequate handling of qualitative data and the need to either develop ways to measure
abstract issues or acknowledge certain issues sit beyond a model's scope.
● Ambitious task of gathering data across the entirety of GB, emphasising the challenges
in spatial mapping of comprehensive data.
● Need to align this discussion with that going on regarding digital twins and data.
Community engagement and trust
● Importance of developers having a clear vision of how tools fit in their processes,
effective communication with stakeholders and training to encourage constructive use of
tools, especially by communities and other stakeholders.
● Understanding what trust means to communities, the significance of robust regulatory
oversight, and the need for developers and utilities to have a clear, concrete vision
aligned with community needs.
● There could be a role for a publicly funded engagement body to build trust (the NFU does
this for landowners).
What’s next for LIPP?
This was the last Panel meeting in the diary for 2023. The Panel will produce a White Paper
summarising its work to date. This will be launched at an event in March 2024 which will also
look forward and consider future opportunities in this area. At the same time, an evaluation of the Panel’s work will take place which will also explore options for future funding and a new ‘home’ for the Panel.
Future funding for the Panel’s work: Ideas included Network Sector Innovation Funding (Ofgem and Ofwat both have relevant funding streams – e.g. NIA - Ofgem’s include community engagement funding for networks). Reference was made to Ofgem's recent decision on local energy institutions and a suggestion was made to leverage the decisions / build on the momentum following Nick Winser’s report and Ofgem’s work on regional energy planning to get funding for new community planning tools that could help deliver against these agendas. A future home for the Panel: Suggestions included the National Infrastructure Commission (NIC), the Royal Town Planning Institute, the Institute of Engineering and Technology and the Institute of Civil Engineering. Some thought that compared to being tied to industry/industry funding this would offer greater independence and transparency. Others thought the industry should fund the work as they have the most to gain from tackling this issue.
The Panel then discussed setting up a Panel working group in 2024 to develop a set of
requirements for the development of stakeholder/community-facing planning tools. The goal would be to produce a ‘How to get an NSIP’ guide/flow chart which could help make the planning system less adversarial and more collaborative.
The working group would include members of the Panel interested in a more granular discussion along with representatives from project promoters and tech developers. The guide/flow cart would be tested by a developer in one of their communities (SSENT have expressed an in-principle willingness to be a test case). The final version would be an open-source set of requirements placed in the public domain.
Many members present expressed an interest in getting involved in the working group. There was a discussion about the benefits of cross-pollination of ideas between sectors, particularly in the energy and water domains, and the need for a policy perspective around the table. The importance of considering the intersection between energy and water, especially in the context of climate resilience and with the evolving landscape of technologies like hydrogen, and the fact that from a community viewpoint, it doesn’t matter which sector is involved, was strongly made. There was also a discussion on the need for background funding to initiate the next phase of this work and to determine an appropriate home for the working group.
The Chair will be in touch with Panel members and relevant stakeholders bilaterally over the
coming weeks to get their steers and in-puts into the final White Paper, March event, future
funding/ home for the Panel and suggested working group on stakeholder/ community tools.
The meeting closed with thanks to all attendees for all their work with the Panel and support over the last year and to Continuum Industries for instigating the process and providing seed corn funding.
Attendee list
Members
Dr Karen Barrass Founder and Director, Climate Insights
Dustin Benton Policy Director, Green Alliance
Sharon Darcy Linear Infrastructure Planning Panel Chair
Aiden Gill Infrastructure Team, National Farmers’ Union
Ada Lee Infrastructure Specialist, Royal Town Planning Institute
Andrew Lovett Professor of Geography, University of East Anglia
Andy Manning Team leader, energy networks and systems, Citizens Advice
Rosie Pearson Chairman Community Planning Alliance and Founder Pylons East
Phil Watson Strategic energy projects manager, Suffolk County Council
Observers
Eric Brown Executive Adviser, Energy Systems Catapult
Lorna Finlayson Head of Networks, Scottish Government
Paul Hickey, Ofwat/RAPID (Regulators’ Alliance for Progressing Infra.
Development)
Ragne Low, Deputy Director, Onshore Electricity, Scottish Government
Faith Pashley Senior Policy Manager, Energy Systems Management and Security,
Ofgem
Margaret Read Director of Policy, National Infrastructure Commission
The following members of the Continuum team will also attend to provide practical support
and observe:
Grzegorz Marecki, CEO, Andreas Malekos, Head of AI, David Sigsworth, Advisor, Charline de Dorlodot, Comms and David Costello, Environment and Consenting
Apologies: Dr Sue Chadwick, Mark Enzer, Alan Farquhar, Richard Benwell, Diarmid Hearns, David Clubb, Karen Alford
The views expressed in this note do not necessarily represent the views of all Panel members or of Continuum Industries.
Comments