Cities launch ‘observatory’ for ethical AI

30 June 2021

by Sarah Wray

Barcelona, London and Amsterdam are leading the Global Observatory on Urban AI project to help cities deploy artificial intelligence effectively and ethically.

Laia Bonet, Barcelona’s Deputy Mayor for the 2030 Agenda, the Digital Transition and International Relations, said it’s important for cities to take a leading role.

“AI systems are riskier for public institutions than for private companies because we have an absolute duty to ensure equal opportunity,” she told Cities Today.

The initiative is part of the Cities Coalition for Digital Rights, which was founded in 2018 by New York, Amsterdam and Barcelona, and now has over 50 members.

Bonet said that AI has the potential to “transform” the way cities deliver services but poor data quality or biased samples could lead to it causing harm and amplify existing inequalities.

Checks and balances

Barcelona is at the early stages of using AI. It recently launched an AI strategy and is piloting applications to monitor beach occupancy levels, classify incident reports from citizens and inform decision-making in social services.

With the right foundations in place, opportunities exist for cities to use AI to become more proactive in providing social services and support.

Keeping up with regulation, development and use cases can be challenging, though, and the Observatory aims to help cities through evidence-based reports, policy standards, and examples of good practice.

Individually, many cities have started to develop and implement their own ethical AI strategies. The Observatory will synthesise and amplify lessons learned to benefit all cities.

Bonet said: “We don’t know whether these checks and balances will be enough to ensure an ethical model. We cannot see the future, but what we can do is to share among cities what we are doing to systematise the impact our strategies are having, and work to create the right support and guidance. That is what the Observatory is about.”

Hidden risks

Facial recognition is often flagged as among the riskiest AI applications and one of the Observatory’s priorities this year is to produce a report on its use in public space, including a set of criteria to assess the ethics of such systems.

Bonet, who is also Chair of Eurocities’ Knowledge Society Forum, has called for an outright, Europe-wide ban on biometric recognition systems, going even further than the European Commission’s proposals.

Beyond facial recognition, though, there are plenty of other algorithmic decisions which could have a huge impact on people’s lives.

“You can imagine a system that systematically rejects applications for municipal assistance because the data it is based on includes mistakes [or because of] profiling a person or a community,” Bonet said. “This actually happens in some countries already.”

Atlas of AI

Recent incidents have highlighted some of the risks. In the UK last summer, the government was forced to back-track on calculating A-level results based on a controversial algorithm after accusations that the system was biased against students from poorer backgrounds. Prime Minister Boris Johnson later called it a “mutant algorithm”. Elsewhere, a “landmark” ruling in the Netherlands found that a predictive AI system for detecting fraud in areas such as social security benefits contravened privacy rights.

In 2021, the Observatory will produce a report on the ethical governance of algorithms in cities. The study will propose an analytical framework to assess policies based on criteria including fairness, explainability and transparency.

The project, which will be delivered in partnership with UN-Habitat and CIDOB-Barcelona Centre for International Affairs, will also launch an online ‘Atlas’ of best practices in urban AI, which will include factsheets on specific applications.

  • Spotonvision
  • Govlaunch

My City, My Life

Valérie Plante, Mayor of Montréal

  • Spotonvision
  • Govlaunch