We launched a series of three public events where local, national and international experts delved into some of the huge questions we must ask ourselves in a world that’s increasingly data-driven.
Algorithms and automated decision-making have been with us for years, especially in industrial systems where a condition can be met with a predetermined response.
Computers are very good at handling these tasks, where the parameters are coded and the outputs of those processes often manifest themselves physically – an aircraft maintains altitude, a biscuit achieves consistent taste, an emergency pumping station pumps water.
But more and more automated decisions are being used to manage social issues, where the output of algorithms can determine if someone gets a home, the right grades or a visa.
- Can we trust automated decision-making systems to make the right decisions?
- How can we ensure that they are doing what they have been designed to do?
- What is the place of these processes in times of crisis?
As part of the ongoing development of the Declaration for Responsible and Intelligent Data Practice, ODM presented Politics, Policy and the Algorithm, a series of discussions that will explore these questions.
Policy and service delivery – 23 February – 6.30pm to 8.00pm (GMT)
The first in the PPA series looked at the use of Artificial Intelligence and automated decision-making systems in the development of policy and service delivery, what they are being used for, what they enable us to do, the decision process that enables these systems to be used, and the challenges faced.
We heard from Tom Forth from The Data City and Tom McNeil, Strategic Advisor to the Police & Crime Commissioner and Ethics Committee member of the West Midlands Police, chaired by Natalie Jameson, CEO of the Heroworx Institute.
Bias, design and implementation – 2 March – 6.30pm to 8.00pm (GMT)
This discussion explored the impact of the bad implementation of Artificial Intelligence and automated decision-making systems, whether these problems are avoidable or if it is an inherent characteristic, especially when these technologies become a manifestation of a broader policy objective, and whether automated assessment processes are intrinsically fair.
We heard from Chai Patel, Legal Policy Director at the Joint Council for the Welfare of Immigrants, Isabel Nisbet, author of ‘Is Assessment Fair?’ and founding CEO of OfQual, Amos Toh, Senior Researcher for AI and Human Rights at Human Rights Watch.
The session was chaired by Jane Eckford, Board Director of Open Data Manchester.
Scrutiny and accountability – 23 March – 6.30pm to 8.00pm (GMT)
This discussion looked at how, if Artificial Intelligence and automated decision-making are going to be implemented, can these systems can be understood and policed. How do we make these systems legible and what are the limits of accountability and liability in ADM systems?
The panel was made up of Jeanette Hofmann, Head of Politics of Digitalisation at the Berlin Social Science Center (WZB), Nicolas Kayser-Bril, Reporter at Algorithm Watch, Keeley Crockett, Professor of Computational Intelligence at Manchester Metropolitan University.
The session was chaired by Peter Wells, former Head of Policy for the Open Data Institute.