• Opinion
  • March 27, 2019
  • 6 minutes
  • 1

AI ethics and Dubai: a city’s attempt to govern the ungovernable?

Opinion: There is no clear path to hard(er) regulation

This piece was written by Andrew Collinge, Advisor, Smart Dubai. For more like this, see our digital government newsfeed.


Last month, Will.I.Am — the musician turned entrepreneur — visited the World Government Summit in Dubai. His calling card? A voice assisted Artificial Intelligence (AI) platform, developed by his own AI company, IAM+. His message? Like this product, broader AI needs to protect the privacy of individuals and increase their control over how their data is used.

Smart Dubai —  the city’s technology and data accelerator — is similarly ambitious and careful. We are clear that exploration of the innovation potential of AI systems is matched by governance that mitigates against the downside risks.

            • Want to write for us? Take a look at Apolitical’s guide for contributors 

We see ethics — being respectful to a core set of values based on fairness, accountability, trust and explainability — as core to society’s adoption of AI and wider market development.

For Smart Dubai, the imperative stems from the fact that much of AI’s transformative impact lies within city services and operations. Sidewalk Toronto is the most obvious current example of how a digitally-rich urban regeneration project can bump into serious public concern about broader data privacy issues. Whether it is commerce, community or government, trust is what makes for strong cities.

And even though we are arguably still in the foothills, AI introduces another dimension of trust in the public sphere. It is vital that government openly acknowledges AI ethics and brokers a multi-stakeholder discussion around them.

This is why Smart Dubai published its AI Ethics Principles and Guidelines. These are helping to develop the “learning-by-doing” and agile governance approaches we believe are needed around the application of algorithms and machine learning.

In December 2018, the European Commission published draft guidelines on ethical AI for consultation (final version due in April 2019). The IEEE Ethics in Action group is producing really high quality work on ethically aligned design of AI. We argue that remote international and national government bodies cannot replicate the proximity and active research and development environment of cities.

The city as an (ethical) AI laboratory

 There are three main elements to our work:

  1. Smart Dubai’s AI Lab — a production line for AI city services, on which the planning, experimentation and stabilisation of use cases takes place. The Lab is also leading on the vital training of public sector staff. 
  2. Ethical AI principles and guidelines — a differentiating factor of our work in a field of other examples is a self-assessment tool, designed to help developers and implementers of AI to judge their systems from various ethical angles. The broader evidence base that this city-level feedback loop generates allows us to further enhance the support we give to Dubai’s AI ecosystem.  
  3. AI Ethics Advisory Board — comprising legal, technology and ethics experts, and representatives from academia and government. Deliberately diverse, this board will bring together an array of perspectives so it can govern and provide strategic direction for the first two elements of our AI work. 

This three-track approach forms a powerful proposition, based around one truth — AI ethics is a field in which no-one has all the answers. Indeed, the diverse competencies and experience needed makes collaboration essential.

Future direction: collaboration, but what about regulation?

The most important note to strike in ending this piece is that, like almost everywhere else, Dubai is in the early stages of sophisticated AI implementation.

In this sense, there is no clear path to hard(er) regulation. Indeed, some have argued that ethical AI principles and guidelines are too broad to be meaningful, and should be dropped in favour of educating public sector staff to exercise good human judgement alongside AI.

We do not necessarily disagree with this point of view. Who knows whether we will be fixating on our own AI Ethics toolkit in three years’ time? We may find plenty of other policy instruments that work better with the development path of this exciting, ambiguous technology.

That said, by publishing our ethical AI principles, guidelines and our voluntary self-assessment tool, Smart Dubai is signalling, first, the importance of ethics to broader adoption, and second, the leadership and partnership working that is needed, if managing the upside and downside risks of AI is to be normalised.

This move towards adaptive and agile governance that enhances, rather than limits, innovation is how we will deliver ethical AI today, and in the longer term, build trust in the technologies driving the digital society and economy of tomorrow. — Andrew Collinge

(Picture credit: Pexels)

Discussion

Leave a Reply

to leave a comment.
Master the skills you need for the public service.

Discover inspiring resources, tools and policies.