Why missing women’s voices at AI conferences should spark concern

Although women make up only 30 per cent of those working in artificial intelligence globally, they are still under-represented at industry conferences
A good mix of people ought to be included in conversations about AI for the technology to be developed in a responsible way that truly serves people

Melissa Zhu

Published: 5:30am, 22 Dec 2023

Why you can trust SCMP

At China’s largest AI event, the World Artificial Intelligence Conference, in July, 20 speakers were scheduled to speak. Only one of them, Advanced Micro Devices (AMD) CEO Lisa Su, was a woman.

This is disappointing but not surprising. Women in technology have long been sidelined, despite their significant contributions to the field.

Earlier this month, The New York Times drew flak for an article listing a Who’s Who of AI that mentioned 12 men and zero women.

With ChatGPT still the hottest topic in AI a year after its launch, it made sense that many of those featured in that article were associated with the chatbot’s developer, OpenAI, including CEO Sam Altman and the billionaires who injected funding into the company. Yet, Mira Murati, OpenAI’s chief technology officer who oversaw the roll-out of ChatGPT and DALL-E, was notably missing.

The list also glaringly omitted Fei-Fei Li, the Stanford scientist often called the “godmother of AI” for creating ImageNet, a large visual database that significantly accelerated the advancement of computer vision applications such as facial recognition, medical imaging and autonomous driving.

At tech conferences, which are important for knowledge sharing, networking and collaboration, women are also conspicuously missing.

In one bizarre fiasco, the DevTernity software conference was cancelled after organiser Eduards Sizov admitted to featuring an “auto-generated” woman, Anna Boyko, in the speaker line-up. He said Boyko was a “demo person” from a test website added to the line-up by mistake.

Mira Murati, chief technology officer of OpenAI, speaks during The Wall Street Journal’s WSJ Tech Live Conference in Laguna Beach, California on October 17, 2023. Photo: AFP

Whether this inclusion of a fake woman was deliberate, the problem runs deeper. All-male panels are so common across different industries that there’s a term for them – “manels”.

As much of the discussion on the topic is US-centric, I decided to see if AI conferences in Asia fared any better. To do this, I manually counted the men and women I could identify in line-ups for all the AI conferences I could find taking place physically in Asia in 2023 and 2024.

The resulting data on 21 events, gathered from Google searches, is by no means comprehensive. I also assumed that all the speakers listed are real people. And it is important to acknowledge a possible margin of error because gender is not always accurately inferred from names or appearances.

Yet even from this small sample, it was clear to me that women are very much under-represented in conversations about AI. Only one out of the 21 events featured anything looking like equal representation – the International Joint Conference on Artificial Intelligence held in Macau in August had nine invited speakers, with women slightly outnumbering men at five to four.

China leads the US in race to curb the dark side of AI

For all the other 20 conferences, women made up less than 30 per cent of the speakers; the figure fell under 15 per cent for more than half the conferences. Two smaller events had no female speakers at all.

The numbers were, at times, staggering. I counted 69 male speakers and 13 women in the line-up for the Teamz Web3 Summit that took place in Tokyo in May. In the list for Big Data & AI World in Singapore in October, I counted 78 men and 11 women.

AI is rapidly reshaping our world, in areas from healthcare, finance and education and national security. While I believe that AI-driven technological advancements could benefit many people, these benefits may not necessarily be evenly distributed. A recent Goldman Sachs report found, for instance, that women are more likely to do jobs that will be negatively impacted by automation.

Even more concerning, AI algorithms trained on data that historically skews white and male have been shown to be biased against women, ethnic minorities and low-income people, with far-reaching implications including wrongful arrests and loss of job opportunities.

It is often women who have drawn attention to these biases. Researchers Joy Buolamwini and Timnit Gebru’s Gender Shades project, for example, led to Microsoft and IBM improving their systems. Scholar Safiya Umoja Noble’s book Algorithms of Oppression draws attention to how search engines like Google may reinforce racism.

In an ideal world, everyone developing AI systems – man or woman – would always be conscious of inclusivity and ethical considerations. But each of us has a different lived experience, and realistically, we are more likely to consider issues that impact us or the people around us.

It is therefore imperative to ensure a good mix of people – in terms of gender, age, race, income and other factors – are involved in conversations surrounding AI.

Women are under-represented in AI, making up about 30 per cent of the field globally according to the World Economic Forum. Even so, many conferences feature a disproportionately small number of female speakers (software developer Aanand Prasad’s Conference Diversity Distribution Calculator is a handy tool).

Open databases of women experts like the directory of Women in AI Ethics could provide a starting point for organisers looking for more diverse speakers. Companies can also actively field female executives to represent them at conferences, and provide them with suitable resources and training to become thought leaders.

05:03

How does China’s AI stack up against ChatGPT?

In a survey of 500 women in the United States and the United Kingdom who had attended tech conferences, three-quarters of respondents said they were more likely to attend a conference that features a woman. Making the extra effort to feature a few more women on panels could encourage more women to attend, be inspired and possibly speak at other events themselves.

Hopefully, in time, there will be enough women speakers at all conferences that no one will have any reason to “auto-generate” one.

Melissa Zhu is a freelance writer and editor who recently completed a Master of Science in Computational and Data Journalism at Cardiff University

Post