AI can have a transformative effect on our cities, our infrastructure, on community life, and on the services and products we receive. The current challenge faced by humanity shows how AI can help model the spread of the Covid-19 pandemic, and consequently, drive effective contact tracing and strategies to direct valuable health kit to where need is greatest. It also illustrates how trade-offs with ethical issues around data and AI need careful consideration too.
If we treat AI carefully and responsibly, holding ourselves to the highest standards of ethics, innovation can be amplified and its potential harms avoided. The value of its use must be determined by the humans developing, operating and using it. Guardrails - in the form of a next generation of ‘enabling’ tools and services – are needed to advance our use and application of AI systems.
We do not start from the ground floor. Smart Dubai’s AI ethics toolkit, released in 2019, includes high level principles and guidelines structured around Fairness, Accountability, Transparency and Explainability, as well as a self-assessment tool that helps developers and commissioners of AI measure ‘ethical performance’ against those guidelines.
Very significant entities – the Road and Transport Authority, Dubai Police, DEWA (Dubai’s main utility provider) – have formally acknowledged their adoption of the self-assessment tool.
A multi-disciplinary AI Ethics advisory board guides our activities in this dynamic space.
Where next – a two-stage commissioning approach with an emphasis on co-production and collaboration?
In 2020, Smart Dubai wants to collaborate to develop new, practical tools and approaches that lower the ‘red flags’ raised by our self-assessment tool. If we are suggesting to an AI developer they should be working harder to explain their system to consumers, we want to provide them, or point them to, the tools to allow them to do so. Our preference is that these tools to be self-service. We think this approach matches market activity and research and development in bias identification, fairness, explainability, privacy, as well as broader audit services.
We understand that we need to familiarise ourselves with existing activity in an emerging market like AI before we move to full commissioning. Furthermore, we are particularly interested in understanding the potential to co-produce and collaborate - i.e. develop and test tools with resources and skills from Smart Dubai (we are amongst the most sophisticated data governors and innovators in the world, we run large parts of the city’s data operations, and through intense private sector engagement can provide access to a local AI ecosystem with whom tools can be trialed and market tested). Our dual position as a direct commissioner of AI and wider city data governor intent on growing AI capacity and maximising its impact means we are keener on this ‘learning together’ approach than simple ‘off the shelf’ commissioning.
Our request for information
Hence an initial Request for Information (RFI). We encourage responses from a range of organisations – e.g. private sector, other public sector, academia, or indeed from specialists in narrower fields such as legal. We understand you are busy and have endeavoured to keep questions to the point. It is up to you to propose tools and solutions in areas other than those we have specified.
Responses will be used in a final Request for Proposals (RFP) which will contain a more detailed set of requirements that will allow us to get the best out of a rapidly developing market, and to offer the best commercial and development opportunities in return.
If you are interested in participating please send an email to c_akmaral.orazaly@smartdubai.ae in order to receive a full copy of the RFI document.