Event round-up: Demystifying urban applications of artificial intelligence

On Thursday 3 June 2021, techUK hosted a short webinar on emerging urban applications of artificial intelligence, exploring how these are driving better outcomes for our people, economy, society, and planet in urban environments.

This session was focused on providing decision-makers with practical advice, use cases and examples that can be leveraged in the real-world, and we’re extremely grateful to our expert panelists and speakers for joining the call!

Up first, techUK welcomed Ben Court, Head of Analytics at Civica, who delivered a powerful presentation on how Civica has leveraged flexible AI capabilities to deliver transformative outcomes in city hospitals around the UK.

During Ben’s presentation, delegates learnt about the AI-based tools that had enabled Civica to run independent tests, produce insights and deliver relevant, cost-saving information to clinicians and hospital decision-makers in the right place, at the right time.

Ben also provided attendees with an in-depth overview of the methods- such as decomposition trees- that were used to communicate insights to key stakeholders. Zooming in on a particular use case from the cardiology department of a UK hospital, Ben demonstrated that AI enabled clinicians and decision-makers to understand where resources could be targeted, driving cost savings, reducing patient waiting lists down, driving re-investment in other areas, and outperforming manual methods of data analysis.

Drawing on this example, Ben raised several key lessons with the audience. Firstly, Ben suggested that it is critical to develop a diverse team with an adequate blend of skills capable of executing a data-centric project of this kind. Secondly, Ben stressed the importance of validating the model used, the decisions supported, and using human-in the loop testing to mitigate unintended consequences related to fairness, equity and bias. Moreover, Ben highlighted that it is not always appropriate to use machine learning, and that the exact nature of the solution used should be proportionate to the problem at hand. In this regard, Ben reiterated the need to develop close relationships between clients and those building AI applications. Finally, Ben presented the notion that AI techniques used to drive cost savings in hospitals could be applied elsewhere, particularly in other areas of urban infrastructure. As such, more thinking was needed around how AI solutions can be repurposed.

“AI enabled clinicians and decision-makers to understand where resources could be targeted, driving cost savings, reducing patient waiting lists down, driving re-investment in other areas, and outperforming manual methods of data analysis”

Next up, Anna Craciun (Partnership Manager at Humanising Autonomy), delivered an insightful presentation highlighting how Humanising Autonomy are developing AI-based, customisable solutions for the UK’s urban mobility sector.

In the opening stages of Anna’s presentation, delegates learnt about some of the core technical approaches underlying Humanising Autonomy’s AI-based product offerings.

For example, our audience learnt that edge capabilities are employed by Humanising Autonomy to enable real-time decision-making and computation in fast-changing urban environments. Similarly, Anna explained that cloud functionalities support interventions with longer decision-support time horizons (such as accident analysis).

Subsequently, Anna raised a number of challenges facing the application of AI in urban systems. First and foremost, Anna highlighted the increased dependency of UK citizens on private transport (for personal travel and deliveries) and loss of trust in public transport that has been triggered by the pandemic. Anna also recognised the difficulty inherent in data quality, and the challenge of visioning insights to users and decision-makers in accessible ways.

Despite these significant challenges, Anna then highlighted how AI-based tools had been employed to enhance city management across the UK, enabling decision makers to understand the position of people, their orientation, and how changes in infrastructure impact the way that people interact with city services and behave in public spaces.

Likewise, Anna explained that customisable AI-tools are being used to support better fleet management in the public transport space, demonstrating that AI is driving levels of safety in public transport use and helping to identify where there are near miss or hard break hotspots. In turn, Anna explained, these hotspots could be used to explore possible interventions and avoid grave situations before they arise.

Lastly, Anna outlined how Humanising Autonomy’s AI-tools had supported decision-makers in dealing with the unprecedented levels of risk and uncertainty driven by the pandemic.  In this domain, Anna identified that AI had enabled decision-makers to understand human behaviour, enhance the efficacy of health restrictions, and enabled critical workers to get where they needed to go.

Here, Anna also touched on the lack of data about marginalized users of transport infrastructure (such as wheelchair users) and suggested that AI-based tools can be extremely insightful to understand what the quick fixes are, as well as, how urban environments can become truly inclusive.

“AI-based tools had been employed to enhance city management across the UK, enabling decision makers to understand the position of people, their orientation, and how changes in infrastructure impact the way that people interact with city services and behave in public spaces”

As Anna closed her presentation, Matt Armstrong-Barnes, Chief Technologist – Artificial Intelligence at HPE, joined the conversation, quizzing our panel and providing his own perspectives on the topic.

In the ensuing live audience participation, our expert speakers discussed common misconceptions of AI- such as the notion that AI is inflexible and difficult to customise. Our panel agreed that AI is, contrary to popular belief, easily configured to bespoke needs and should be viewed as a decision-making tool that can be co-designed with clients and end-users from the outset.

In a similar vein, our panel suggested that those interested in adopting urban AI solutions should “pick a tool to the problem, rather than fitting a tool the problem” and that failure is an option when it comes to scoping out the efficacy and proportionality of AI solutions during the early stages of adoption.

Building on these points, our panel indicated that those considering AI adoption should begin with the problem at hand, understand existing business processes, and engage with those individuals that need help with decision-making.

Using this understanding and consensus that AI is the right tool, our panelists suggested that it is then important to demystify the AI tools further, ensuring that decision-makers across an organisation can play a part in shaping and using the tools in an accessible way.

Finally, our panel discussed some of the central challenges facing AI adoption in an urban context. Here, our panel suggested that replicability is a significant challenge. While a solution may work effectively in one domain, there is a need to ensure that the solutions are durable and transportable in other sectors and problem spaces. Additionally, the panel explored the degree to which AI is disrupting long-established, manual processes of data analysis.

In this segment, our panel reiterated that AI should be viewed as an enabling tool that helps human decision-makers and analysts to focus their efforts around business critical insight areas.

In short, the session was incredibly wide-reaching and techUK is extremely grateful for our panelists and those that took the time to attend. If you would like to find out more about the work that techUK is doing in this space, via our Smart Cities Working Group, please feel free to get in touch with [email protected] today!