10 Reflections from Smart City Expo 2023 in Barcelona
LOTI organised a delegation to attend the Smart City Expo World Congress (SCEWC) in Barcelona, the SCEWC is the world’s biggest event on urban innovation. Attending as part of the LOTI delegation, here are my 10 reflections from the event.
1. More focus on outcomes and less on shiny tech
The smart city world has often been guilty of a kind of valorisation of technology over all else. Selling expensive technology against a background of hype and PR which often failed to address any real problems. For this reason it was refreshing to see an agenda that centred around the challenges cities are trying to address including those aligned with LOTIs current strategic focuses climate change, social care and housing. Beyond this, urban mobility, planning, energy infrastructure and construction also featured as contexts for deploying smart city technology. Notably inclusion and ethics featured in the main programme including in an insightful keynote on how to evaluate for bias in machine learning and AI modelling given by the Chief Ethicist at Hugging Face (more on this later).
2. Getting back to Nature – techno determinism is going out of fashion
The conference programme and exhibition featured a noticeable shift in tone away from tech-utopianism towards a more reflective position on how the digital age has created a degree of alienation of humans from nature. This took the form of pitches on the possibilities that are emerging at the intersection of biology and quantum computing, using plants as switches and stores of energy, discussion of paring back smart city design to make it less digital and more in equilibrium with nature and a subsidiary of Samsung selling an all in one smart city solution that cryptically promises to ‘revitalise the human’.
3. Climate Assemblies and deliberative decision making as part of smart cities
Many cities have assembled citizen assemblies as part of their response to climate change. The most successful initiatives have engaged in participatory budgeting exercises and in Bologna the city governments now have a permanent climate assembly that helps to make policy decisions. The climate assembly has kept the city accountable to its climate goals and maintains public trust in the process. To me, a really smart city engages effectively with citizens to deliver on big challenges like climate change and mitigation.
4. Digital Twin: top down or bottom up?
There is a discussion on how to deliver digital twins in cities. One camp believes that a top down data platform is the answer and city states such as Singapore have evidenced how to make this work. In London we are in the other camp. For a city which is effectively 33 cities plus the Mayor’s office, the transport authority and the fire, police and health services we need to work more iteratively. Creating data models around narrow usecases such as traffic modelling, as is being done at TfL, is the first step in our journey towards more sophisticated and integrated digital twins. This enables us to build capability and buy in and keep focused around delivering on outcomes for the city rather than implementing tech for techs sake. I hope that in the future, the work LOTI is doing with London Councils and TfL to create a EV Chargepoint Utilisation map will one day pivot into becoming the data source for a kerb side digital twin, which in turn might share data with the TfL traffic modelling digital twin.
5. Finding the golden usecase for apps to drive adoption
Everyone wants an app but hardly anyone wants to install new apps! Local government technology teams are often asked to build new apps for a narrow use case that, if delivered, rarely build a user base beyond a small group of dedicated users. The City of Kyiv has reversed this trend with their city app which is used by 85%+ of the population. They have integrated transport, safety and e-democracy in one app which has offered democratic resilience in a time of war. The application also provides notifications on air raids and floods which is a significant driver of user adoption. While thankfully most cities do not need to offer this functionality, the lesson to learn here is that delivering core functionality with universal appeal first builds user adoption and grows a much bigger audience for public engagement like polling, consultations and even direct democracy. The democratic resilience this offers could be essential in situations such as future pandemics.
6. AI Sustainability
Generative AI was a big topic at this year’s conference and one of the issues raised was the high costs of training and operating AI large language models (LLMs) such as those that underpin Chat GPT. It is estimated that training GPT 4 cost over $100 million dollars in energy cost. This is the energy required to operate the data centres and power the computational processing to create the model. This doesn’t include ongoing energy requirements that come with the API calls required to operate front ends such as Chat GPT. This means we need to think about how and why we use these models and how we power them. Part of this will be selecting the appropriate size model for the task in hand while encouraging those developing and training LLMs to do so using renewable power.
7. Horizon Europe is back in play for London
The UK has come to an agreement to continue to participate in the EU Horizon Europe research and innovation funding programme. While the UK technically never left the programme, there was reluctance to enter funding bids with the uncertainty of the UK potentially exiting the programme in the near future. With this resolved there is an opportunity for London Boroughs and partners organisations to bid for funding for smart city initiatives in the vein of previously successful work of the Sharing Cities programme that received funding from Horizon 2020.
8. Ethics – how to evaluate machine learning and AI models for equity
Margaret Mitchell’s keynote on evaluating bias in AI Models was a highlight of the conference. It was simultaneously technical and applicable while being broadly insightful and relevant to ethical thinking across smart cities more generally. Margaret outlined an approach for practically checking for bias in AI and Machine Learning models that involves setting evaluation criteria based on the risk faced by specific vulnerable groups in context. You can find more details on this approach in on the Hugging Face website.
9. Quantum will be the next hype cycle trend
Quantum was being mentioned all over the conference. From computing to materials. There were theoretical propositions on the modelling of the quantum effect on cells of electromagnetic fields that could lead to devices which can heal skin with radio waves and more grounded discussion on issues with encryption that will become very real once quantum computers are available outside the lab. It will be the next big hype cycle trend in technology and data.
10. Post scarcity
The digital revolution and the advent of 3D printing was meant to be the end of the post scarcity world. With the marginal cost of production of digital goods falling to near zero and the theoretical ability of 3D printers to print themselves posing a game changing proposition for our current political economy. It hasn’t quite worked out that way but one presenter offered up the suggestion that LLMs might again offer an opportunity for another world. We already know that LLMs offer the opportunity for very bespoke design, with LLMs able to write instructions for 3D printers perhaps we can start to imagine a future where highly bespoke creations can designed and produced at very low cost?
Jay Saggar