The Legal Challenges Facing Algorithmic Decision-Making in Kenya
DOI:
https://doi.org/10.66563/6hwtwb33Abstract
This paper evaluates the existing non-discriminatory legal framework in Kenya to assess its limitations concerning the increasing use of computer algorithms in decision-making. It argues that the legal tools available may not be adequate to address the challenges of automated decisions. The detection of algorithmic discrimination is a huge challenge that the law does not anticipate. This is because evidence of discrimination is not apparent due to the technical and opaque nature of Artificial Intelligence (AI) and Machine Learning (ML) processes. Furthermore, when evidence is obtained, interpreting it requires highly skilled technical expertise which may not be available. The paper further argues that while the concept of indirect discrimination may be useful in addressing algorithmic discrimination, it may not be enough: additional actions may be required such as the imposition of a requirement for interpretability of algorithmic systems by law. At the same time, whereas the Data Protection Act, 2019 (DPA) contains some progressive provisions in relation to automated decisions, it has its challenges such as compliance and enforcement, and the fact that the DPA only applies to personal data. This paper shall then conclude by calling for the enactment of a comprehensive non-discriminatory statute that addresses, inter alia, the unique challenges of algorithmic discrimination.
References
Published
Issue
Section
License
© Copyright
University of Nairobi Law Journal. All rights reserved. No part of this publication may be reproduced, stored in retrieval system or transmitted in any form or by any means, electronic, mechanical or photocopying, recording, or otherwise without the prior permission of the University of Nairobi Law Journal.
The University of Nairobi Law Journal believes in the strength of due diligence exercised, that this work does not contain any material that is the subject of copyright held by another person. In the alternative, they believe that any protected pre-existing material that may be comprised in it has been used with appropriate authority or has been used in such circumstances that make such use permissible under the law.
Although every care is taken to ensure the accuracy of this publication, supplements, updates and replacement material, the authors, editors, publishers and printers do not accept responsibility for any act, omission, loss, damage or the consequences thereof occasioned by a reliance by any person upon the contents hereof.
