It is a fact that on the 21st century we are depending more and more on the internet to find information. That is why the 21st century can be called the age of information. It can be said that Artificial Intelligence (AI) is highly affected by societies’ norms. One can say that the AI constitutes a reflection of society. So, if a society has discriminatory norms, the AI will be trained upon that. For example, a medical school in the 1980s used a biased discriminatory system to select students. The system was trained to discriminate students coming from an immigrant background and also women. This means that those who trained the algorithmic system on biased data, were biased against people with an immigrant background and against women.
Discrimination can be defined as the unfair and unjust treatment of people based on certain characteristics. One of the Core UN Human Rights Conventions clearly prohibits States from engaging in any forms of racial discrimination. Racial Discrimination according to Article 1 (1) of the International Convention on the Elimination of All Forms of Racial Discrimination is any distinction, exclusion, restriction or preference based on race, colour, descent, or national or ethnic origin which has the purpose or effect of nullifying or impairing the recognition, enjoyment or exercise, on an equal footing, of human rights and fundamental freedoms in the political, economic, social, cultural or any other field of public life. According to Article 2 (1) “States Parties condemn racial discrimination and undertake to pursue by all appropriate means and without delay a policy of eliminating racial discrimination in all its forms and promoting understanding among all races…”.
Moreover, it has been noticed that AI commercials that advertise jobs with high salaries target more male audience than female audience. This is an example of discriminatory use of AI. The reason behind this incident was the fact that males constitute the majority of persons who purchase high salaries. The worst part was that women did not even realise that they were discriminated. The opaqueness of the AI systems makes it very difficult to discover discrimination and its cause. But the law is clear.
States must prohibit all forms of discrimination against women. So, the use of Artificial Intelligence must comply with the core UN Human Rights Conventions on the prohibition of discrimination of women. In order to avoid discrimination in Artificial Intelligence, societies are the ones that need to eradicate discrimination first. Another example of discriminatory effect of AI was the fact that when users of google searched for black teenagers it showed people with mugshots. On the contrary, when people searched for white teenagers, Google images showed pictures of ‘’happy white kids’’. Google responded ‘Our image search results are a reflection of content from across the web, including the frequency with which types of images appear and the way they’re described online’. Google’s algorithmic system reflected societies’ norms. Undoubtably these images could influence people’s beliefs.
Moreover, the use of Artificial Intelligence can be very discriminative when it targets digital groups based on religion, sexual orientation and gender. For example, the Dutch Data Protection Authority found that Facebook targeted people based on sensitive characteristics such as sexual interests, homosexuality, Islam and reproductive health. Due to the rise of terrorism, many migrants from Muslim backgrounds have faced unfair and unjust discriminatory behavior. Angwin and Perris at ProPublica proved that Facebook’s advertisers exclude users by race, such as Black or Hispanic, from seeing different ads.
The above situations are clearly discriminatory, and they violate non-discriminatory laws. All of the above constitute the disadvantages of Artificial Intelligence. It is up to us then to make the use of AI fairer and non-discriminatory. A way to do that is to train the AI to comply with equality laws and more specifically with Human Rights Law. As it is mentioned above, societies first need to limit discrimination in order for the AI to be trained and become more fair.
Contact us at info@aapartners.law or 22250597.
S. ANTONIOU & ASSOCIATES LLC is a lawyer’s liability company founded by Stavriana Antoniou.
© S. ANTONIOU & ASSOCIATES LLC. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of S. ANTONIOU & ASSOCIATES LLC. The information provided in this publication does not constitute legal, tax or investment advice and no responsibility is accepted for any loss occasioned directly or indirectly as a result of persons acting, or refraining from acting, wholly or partially in reliance upon it.
