Issue link: https://maltatoday.uberflip.com/i/1266179
3 LETTERS & EDITORIAL maltatoday | SUNDAY • 5 JULY 2020 Mikiel Galea Letters & Clarifications Gender equality in AI THE Advisory Committee on Equal Opportunities for Women and Men recently published an Opinion on the opportunities and challenges for gender equality in relation to Artificial Intelli- gence (AI). It is becoming increasingly urgent to discuss and address AI's im- pact on equality since there is already a widespread use of AI in all spheres of life, and this use is bound to increase. For this reason, the National Com- mission for the Promotion of Equality (NCPE) organised in March the first conference in Malta about AI and its impact on gender equality in spheres such as employment and service-provi- sion. The Opinion recognises that technolog- ical advances such as AI, automation and robotics, are having profound effects on gender equality and women's rights. AI can mirror human biases and even rein- force them, either because biased data is programmed into the algorithm - which is a procedure for solving a mathemati- cal problem - or because biased people create biased algorithms. Additionally, the automation of the decision-making process in AI makes it harder for those impacted by the decisions to understand how these are taken and to identify dis- crimination. We should also be aware of the oppor- tunities AI offers for the development of positive algorithms that can identify discrimination or detect harmful online content against women and girls. A concern for equality should be inte- grated in all phases of AI development. Gender impact assessments should be carried out to analyse whether specific groups of women and men are, directly or indirectly, affected during all stag- es of the algorithmic decision-making. The Advisory Committee also proposes a pro-active way of tackling discrimina- tion, through the use of AI itself, by de- veloping algorithms capable of address- ing existing digital gender biases and prejudice. AI user-friendliness, transparency and non-discriminatory use can be enhanced if diverse groups of citizens participate in its development. However, only 17% of current ICT specialists in the EU are women. We need to eliminate gender stereotypes in education and take posi- tive action to attract, from a young age, more women to STEM fields and, par- ticularly, to AI. An important component in the de- velopment of AI is the transference of research. A multi-disciplinary approach to AI research – one which incorporates humanities, social sciences, technologi- cal studies and gender research – is rec- ommended by the Advisory Committee, since this approach can significantly con- tribute to preventing gender biases from finding their way into algorithms. The Advisory Committee highlights the possibility of establishing an AI reg- ulatory framework through auditing and monitoring AI systems and databases for equality, fairness, accountability and robustness. Such mechanism could be complemented by the implementation of a Code of Ethics for the development, ap- plication and use of AI to ensure that the principles of a human-centric approach, diversity and human rights are safeguard- ed throughout. Capacity-building and allocation of resources for stakeholders, including equality bodies, to be better equipped with knowledge and tools that detect and address discrimination in the context of AI is also recommended. With recruitment being one of the most sensitive areas where AI-related gender biases and discriminatory prac- tices can occur, the Advisory Commit- tee underlines the importance of raising awareness in the HR sector about the use of AI tools, their functioning and potential dangers. Moreover, it is stated that transparency concerning the criteria used in the recruitment process should always be ensured. Knowingly or unwittingly, we use AI on a daily basis and decisions about our lives are regularly being taken or influenced by machines. In order to avoid exacer- bating discrimination across all spheres of society with the spread of AI, we need to be attentive to the impact of AI on equality and develop AI systems that are free from biases and stereotypes. Maria Camilleri NCPE