One early morning last month, as giggling Swiss schoolchildren chased each other along the streets of the cosmopolitan Varembé neighborhood, a line of serious-looking adults in sharp suits and dresses filed into the Centre de conférence Genève. The more than 2,000 participants from over 120 countries were here for the AI for Good Global Summit in Geneva 2019 — four days of presentations and discussions on how to make the world a better place for children to grow up in by leveraging the emerging power of machine learning.
In 2015 the United Nations introduced its “17 Sustainable Development Goals (SDGs)” and the focus here was on how to most effectively direct AI research, development and deployment to achieve these SDGs by the target date of 2030.
The UN’s ITU (International Telecommunication Union) has been organizing the AI for Good Global Summit for three years, drawing a for a multi-stakeholder and inter-disciplinary audience which includes government, industry professionals, UN agencies, civil society, international organizations, and academia.
“This summit is the leading United Nations platform for dialogue on artificial intelligence. AI is being used to fight hunger, mitigate the climate crisis, or facilitate the transition to smart sustainable cities,” said ITU Secretary-General Houlin Zhao in his keynote. “The path to a transformative but also a safe, trusted and inclusive AI will require unprecedented collaboration between government, industry, academia and civil society.” Zhou’s speech reviewed the 17 SDGs, adding AI-specific strategies:
- No poverty: Bring advances in AI to the most vulnerable to ensure algorithmic equity
- Zero Hunger: Use AI to improve agriculture for better food security
- Good Health and Well Being: Standardize a framework for the performance benchmarking of ‘AI for Health’ algorithms to address health issues
- Quality Education: Education can enable students to thrive in a world increasingly augmented by AI-powered technologies
- Gender Equality: Actionable and thoughtful ways to identify and address gender inequalities
- Clean Water and Sanitation: Tackle challenges such as lack of expertise, climate change, resource optimization, and consumer trust
- Affordable and Clean Energy: Improve photovoltaic energy capture
- Decent Work and Economic Growth: Industrial AI to accelerate economic growth
- Industry Innovation and Infrastructure: Bridge the digital divide and increase connectivity gains
- Reduced Inequality: Align all stakeholders in civil society and improve transparency
- Sustainable Cities and Communities: More resilient, innovative systems in smart cities
- Responsible Consumption and Production: Promote the use of AI and other frontier technologies such as IoTs, Big Data analytics, and 5G
- Climate Action: Take advantage of frontier technologies in combating climate change and achieving a circular economy
- Life Below Water: Advance deep sea technologies for autonomous, fast, high-resolution ocean exploration.
- Life on Land: AI-powered drones to help fight ocean plastic and poachers
- Peace, Justice and Strong Institutions: Digital transformation to ensure connectivity
- Partnerships with Goals: Bring diverse stakeholders from around the world together and develop skills for the digital economy and society
One of the buzzwords at this year’s summit was “gender equality,” regarded not only as a fundamental human right but also a critical foundation for a sustainable and peaceful world. The Gender Equality SDG aims to empower women and girls. Although some forms of discrimination against females are diminishing, according to The Global Gender Gap Report 2018 by World Economic Forum, only 22 percent of AI professionals globally are female, while 78 percent are male.
Because algorithms learn from real-world data, AI can potentially adopt and reinforce existing social biases. Developers could for example unconsciously integrate such gender biases into their AI systems.
A recent UN report, I’d blush if I could: closing gender divides in digital skills through education, argues that AI digital assistants with female voices can reinforce existing gender biases. The report notes that although although female voices are increasingly popular in intelligent machines, this was not always the case: “Perhaps the closest relative to today’s all-purpose virtual assistants were speaking car navigation systems. The voices for these systems gave terse, authoritative directions (‘turn left in one block’, ‘go straight for 500 metres’) and were almost always male.”
The trend toward female voiced virtual assistants, the report suggests, “seems to have less to do with sound, tone, syntax, and cadence, than an association with assistance.”
Gender bias can also be reflected in recruiting tools, search engines, and in face recognition systems — which tend to have more errors with female faces. From the MIT and Microsoft collaboration Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification: “We evaluate 3 commercial gender classification systems using our dataset and show that darker-skinned females are the most misclassified group (with error rates of up to 34.7%).”
The AI and ML community are increasingly aware of the biases and at the AI for Good Global Summit a number of actionable and thoughtful ways to identify and address inequalities were discussed.
As Synced previously reported, last year the Conference on Neural Information Processing Systems changed its official acronym from “NIPS” to “NeurIPS” after John Hopkins University professors and students complained the former could create a “hostile environment” due to the term being “vulnerable to sexual puns.”
As societies continue adopting AI applications at an unprecedented rate, it’s believed that now is the time to address gender equality issues. As noted in the I’d blush if I could… report, “Diverse and gender-equal technical teams are urgently needed at a moment when processes to teach and give expression to intelligent machines are being cast.”
The AI for Good Global Summit in Geneva 2019 ran From May 28 to 31. The next gathering is scheduled for May 4-8, 2020.
Journalist: Fangyu Cai | Editor: Michael Sarazen