MONTRÉAL, the 26 nov. 2024 /CNW/ – Mila et CEIMIAtwo global leaders in the responsible development of artificial intelligence (AI), today released the most comprehensive report to date on gender equality and diversity in AI . Drawing on extensive consultations with a wide range of voices and experts, this report and accompanying guide offer decision-makers clear policy recommendations to promote greater inclusiveness in technological development.
The report, an initiative of the AI Working Group responsible for Global Partnership on Artificial Intelligencewas presented today in Montreal, at Canadaas part of an event which brought together the President of the Treasury Board of Canada, Anita Anandas well as world leaders, policymakers and representatives of the AI community.
The report, titled « Towards real equality in artificial intelligence: A transformative AI policy for gender equality and diversity »provides decision-makers and all stakeholders with guidelines for creating inclusive, equitable and just AI ecosystems that promote economic and social development. The report explores best practices, offers policy perspectives and provides recommendations for improving gender equality and diversity in AI and developing guidelines.
Rapid advances in AI are transforming industries and driving economic growth, offering great potential to improve the quality and conditions of life around the world. However, they risk exacerbating existing inequalities by reflecting and amplifying societal biases, discrimination and stereotypes, particularly those affecting women and other historically marginalized groups. There are already documented cases where AI systems and processes have, for example, led to the distribution of non-consensual personal images and online harassment. Lack of diversity in AI ecosystems can also reinforce stereotypes and discrimination by excluding certain perspectives. This has included cases where marginalized groups have had their employment opportunities limited and resources have been distributed inequitably, particularly towards people with disabilities.
“This report is essential because the recommendations address the root causes of inequality with a view to achieving real equality in AI and beyond. Transformative changes in the development and implementation of policies surrounding AI will advance human rights,” says Paola Ricaurte Quijiano, full professor at Tecnológico de Monterrey, associate at the Berkman Klein Center for Internet & Society at university Harvard and co-responsible for the report. “We need to create a strong regulatory framework to limit harmful effects and ensure beneficial development of AI systems and processes for all. »
The main recommendations are classified into four categories:
- Encouraging inclusive design and democratic innovation : integrate affirmative action and institutional inclusion measures, and support the design of inclusive technologies.
- Strengthening meaningful participation in AI governance : promote the active participation of marginalized groups in AI governance in order to ensure better supervision of AI for all.
- Ensure transparency and accountability in preventing adverse effects: establish safeguards and accountability mechanisms among all AI stakeholders to prevent harm and ensure fairness.
- Ensuring adequate access to the justice system : Implement measures to ensure that marginalized groups have access to legal recourse against AI-related discrimination and bias.
The report recommends, among other things, directing funding towards initiatives aligned with inclusive technology design principles, promoting equitable applications, practices and processes in AI ecosystems. The recommendations also focus on developing strategies to enable the public and private sectors to carry out fundamental rights impact assessments before deploying AI systems. Other examples include ensuring algorithmic transparency, to allow those affected by an AI system to challenge its results based on clear and understandable information. In addition, it is proposed to reduce the burden of proof for plaintiffs, thus facilitating access to legal remedies against discrimination and bias generated by AI.
“The establishment of a rigorous regulatory framework is necessary and urgent to prevent harm and work towards fairness and diversity in AI ecosystems which, ultimately, directly impact humans,” adds Benjamin Prud‘man, Mila’s vice president of public policy, security and global affairs and co-lead of the report. “Through inclusive AI policies, we can improve the quality, usability and effectiveness of AI systems, for a more equitable, sustainable and prosperous future for all. »
Alongside the report, CEIMIA and Mila are also publishing a companion guide to help policymakers implement the recommendations. This guide offers a detailed method for building more inclusive AI ecosystems and concrete steps to facilitate this essential industrial transition.
Based on a human rights approach to AI, more than 200 participants, representing more than 50 countries and from diverse backgrounds, took part in consultations or consultative exercises to create the report.
About CEIMIA
CEIMIA is a key player in the responsible development and adoption of artificial intelligence, based on principles of ethics, human rights, inclusion, diversity, innovation and economic growth . CEIMIA develops and implements high-impact applied projects in the field of responsible artificial intelligence. As the center of expertise for the Global Partnership on Artificial Intelligence (GPAI), CEIMIA works closely with 29 governments and more than 150 multi-stakeholder international experts to address global challenges and priorities by supporting the development of applied solutions in responsible artificial intelligence. Our role is to act as a focal point for international partnerships to encourage the development and deployment of such initiatives for the public good. For more information, visit https://ceimia.org/fr/
About Mila
Founded by Professor Yoshua Bengio de The University of Montreal, Mila is the largest university research center in deep learning in the world, bringing together more than 1,200 researchers specializing in the field of machine learning. Based in Montreal and funded in part by the government of Canada as part of the Pan-Canadian AI Strategy, Mila’s mission is to be a global hub of scientific advances that inspires innovation and the development of AI for the benefit of all. Mila is a non-profit organization recognized globally for its significant contributions to the field of deep learning, particularly in the areas of language modeling, machine translation, object recognition, and generative models. To learn more, visit mila.quebec.
SOURCE Mila – Quebec Institute of Artificial Intelligence
Source: Mila & CEIMIA ; Contact: Eric Aach, [email protected]+1 514-569-3594; or Gabrielle Landry, [email protected]+1 514-688-5837
