Algorithmic Discrimination
How algorithms expose the limits of the ICCPR’s anti-discrimination framework
Joseph Higgitt

In recent decades, algorithms have become an unavoidable part of society. They are used to
filter job applications, target police resources and identify terror threats. Whilst they
undoubtedly carry great benefit, algorithms also raise serious human rights concerns, as
improper use can lead to certain groups being disproportionately overlooked by employers, or
minorities being over-policed by anti-terror authorities and police.
The International Covenant on Civil and Political Rights (ICCPR) nominally prohibits
discrimination. However, algorithmic discrimination continues to challenge the ICCPR’s anti-
discrimination rules as, unlike other forms of discrimination, it often lacks the
‘recognisability’ essential to engaging ICCPR rights. This essay argues that the nature and
use of algorithms give rise to three types of ‘opacity’ that make their decision-making process
incomprehensible to most people, meaning that any potentially discriminatory decisions they
make are unrecognisable and thus difficult to contest and rectify. This essay concludes by
suggesting that a procedural ‘right to explanation’ that operationalises existing ICCPR rights
may be necessary to prevent algorithmic discrimination.


Comments