UNESCO hails ‘outstanding’ AI smart city research of Leeds professor

A University of Leeds research project on AI’s role in sustainable smart cities has been selected as “outstanding” by UNESCO.

The work of Dr Evangelos Pournaras was given the highest acclaim in the International Research Centre on Artificial Intelligence’s (IRCAI’s) Global Top 100, an annual list of the top projects that solve problems related to the 17 United Nations Sustainable Development Goals. 

His top-ten project, ‘Collective Learning: Human-machine Collective Intelligence at Smart City Scale’, proposes a novel open-source AI solution to digitally assist and coordinate collective human decisions in smart cities. 

Dr Pournaras is a UKRI Future Leaders Fellow and an Associate Professor at the University’s School of Computing. He uses AI to find a solution to long-standing problems such as power blackouts, traffic jams, and over-crowded parking spaces, which “have unprecedented impact on environment and society”. 

These are often a result of “coordination deficit”, caused by the complexity of human collective choices in aligning power consumption, traffic routes or parking space choices when they are critically needed.  

His work focuses on two goals: 

  • Producing a scalable, automated means for citizens to exchange timely data to create more sustainable collective arrangements of their resources for the environment; 

  • Offering a low-cost, fair, and inclusive solution that is also trustworthy. 

Dr Pournaras, whose research is within a £1.4 million UKRI Future Leaders Fellowship and a network of prominent partners in industry and cities, uses collective learning through digital assistants on devices such as smartphones to deliver recommendations calculated through a decentralised, privacy-preserving network. 

Two visualisations of maps showing congested and optimised movement within a map.

Dr Pournaras' visualisations show the difference between congested (left) and AI-optimised movement within a smart city map.

Collective learning would then be able to optimise decisions at a granular level, such as recommending the right moment to turn off a home appliance to reduce energy, a low-carbon slot to charge an electric vehicle, or finding the nearest bike sharing station with the lowest cost.  

“What makes the difference in these optimised decisions altogether is that they can load-balance and improve the efficiency, while meeting a broad spectrum of sustainable development goals,” the research outline states. “Collective learning comes in stark contrast to mainstream supervised AI solutions that require training with massive personal data and cannot easily scale.” 

Dr Pournaras gives a three-minute precis of his project at the 8th Multi-Stakeholder Forum on Science, Technology and Innovation, which you can watch on YouTube (from 55:30)

Further information