A Review of the ICO’s Draft Guidance on the AI Auditing Framework

15 Pages Posted: 25 Jun 2020

See all articles by Emre Kazim

Emre Kazim

University College London

Adriano Koshiyama

Department of Computer Science, University College London

Date Written: May 12, 2020

Abstract

The Information Commissioner's Office (ICO) timely call for consultation regarding their ‘Guidance on the AI auditing framework: Draft guidance for consultation’ (February 2020) is part of a growing literature concerning the governance of Artificial Intelligence (AI) systems. The ICO’s draft leads the national conversation by producing guidance that encompasses both technical (ex. system impact assessments) and non-technical (ex. human oversight) components to governance and represents a significant milestone in the movement towards standardising AI governance. Welcoming this crucial intervention, we summarise and critically evaluated each section of the draft guidance, offering feed-back in line with the call for consultation. We conclude with a note on what we anticipate will be future debates and by presenting our general recommendations. These are namely:

Data and AI: explicit discussion of the relationship between data protection and how it translates into auditing of AI systems.

Case studies: templates development produced through discussion in an open forum for relevant stakeholders.

Risks of other Machine Learning (ML) systems: in addition to guidance on Supervised Learning further guidance would be welcome about risks presented in other forms of ML.

Regression and Forecasting: discuss the metrics and methods used when an AI system is used to tackle a Regression or a Forecasting problem.

Target audience: better specification (Data Scientists, DPOs, etc.) or a framework should be created where each group is targeted within a structure that integrates their respective duties.

Suggested Citation

Kazim, Emre and Koshiyama, Adriano, A Review of the ICO’s Draft Guidance on the AI Auditing Framework (May 12, 2020). Available at SSRN: https://ssrn.com/abstract=3599226 or http://dx.doi.org/10.2139/ssrn.3599226

Emre Kazim (Contact Author)

University College London ( email )

United Kingdom

Adriano Koshiyama

Department of Computer Science, University College London ( email )

Gower Street
London, London WC1E 6BT
United Kingdom

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
179
Abstract Views
753
rank
196,046
PlumX Metrics