When Tech Meets Women's Rights

Transcript

I can see that some of you are already paying your respects in the chat section. I also pay my respects to the Elders of those traditional lands on which you are joining us from, and I'd like to extend a warm welcome to any Aboriginal and Torres Strait Islander people who may be joining us tonight for this presentation.

So welcome, everyone. I know we have a really large number of Brennan students, alumni and staff here tonight, which is really fabulous to see. I imagine, like me, you're very excited to hear Ramona's incredibly interesting and topical talk. While I have the pleasure of being one of Ramona's colleagues, it's actually rare for me to get a chance to hear her talk about her work at length, so this is a real pleasure for me this evening.

Before I formally introduce Ramona, I'm just going to remind you about a few Zoom housekeeping matters. Please remember to mute your microphone if you are not speaking. If you feel comfortable and your bandwidth allows, please feel free to put your camera on—it can make it a little bit more interactive, and it's nice to see some friendly faces. If you do have your camera on and it starts to freeze, just turn your camera off and it will free up some bandwidth for you.

We will have time for questions at the end, which you can either ask in person or put in the chat. If you want to put your question in the chat, you can start to do that while Ramona is talking so you don't forget, and we will read it out later on at the end.

In terms of the five Reflection on Justice points for the Brennan students who are participating tonight, you will automatically be awarded these if you have joined the session with the name that you have in CareerHub in the university systems. If you have joined the Zoom event using another name, like a nickname, or perhaps you've used someone else's computer, please just put your full name that's in CareerHub into the chat section, and then Crystal and other members of the team can mark you off.

So now it's my great pleasure to introduce you to our special guest for tonight, Associate Professor Ramona Vijeyarasa. Ramona joined UTS in 2017 as a Chancellor's Postdoctoral Fellow, having completed her PhD at UTS and a Master of Laws at New York University Law School. She's an award-winning scholar and currently leads the JD program here at UTS. She's the chief investigator behind the Gender Legislative Index, an innovative tool to rank and score legislation against global standards for women's rights. If you haven't looked at the Gender Legislative Index, please just Google it and it will be the first thing that comes up—it's incredibly exciting to have a look at. This work has been instrumental in the establishment of the Tasmanian Legislative Council's Gender and Equality Audit Committee. Ramona has advised the Australian government and lots of NGOs, both in Australia and overseas, on women's rights and gender-related interventions. Before joining UTS, Ramona held several positions on women's rights in international organisations, non-profits and international NGOs. This work included advancing anti-trafficking victim reintegration networks in Vietnam and Ukraine, filing briefs before the European Court of Human Rights, the Supreme Court of Moldova and the Supreme Court of the Philippines, making submissions to UN treaty bodies, and overseeing research and programming on women and youth in urban spaces. This incredibly rich and varied experience informs Ramona's impact-driven approach to research. She's widely published—last year she published "The Woman President", which explores the difference women leaders can make on women's lives through the law. She's also published extensively on trafficking, modern slavery, work and supply chains, gender-responsive legislation and much more. She's also won multiple awards, including the 2023 American Society of International Law Scholarship Prize in recognition of excellence in international law scholarship involving women and girls, gender, and feminist approaches, and in 2022 she won the Women in AI Award for the Law category. So please join me in welcoming Ramona, who will be speaking about "When Tech Meets Women's Rights". Thank you.

Ramona: Thank you so much, Jane, for that really kind and actually very humbling introduction. It is an absolute pleasure to deliver this year's fifth Brennan Justice Talk for 2023, and also the first for UTS Tech and Social Justice Week. Before I share my remarks, however, I would also like to acknowledge the traditional owners of the land where I stand, the Gadigal people of the Eora Nation, where I'm privileged to live and work every day.

As you'll hear from my talk, I actually spent quite a bit of time living overseas working as a women's rights lawyer. Before I knew it, 11 years had passed. When I moved back to Sydney, I was very positively surprised to see how commonplace it had become to acknowledge the traditional owners whose lands we occupy. This was not something I saw before I left Sydney in 2006. Today, I think it's not just normal practice but also a much more considered and thoughtful one. So I would add, in paying my respects to Elders past, present and emerging and any First Nations people joining us today, that I also take it as a reminder that change, however slow, is always possible.

It was around 12 or 13 years ago that I found myself living outside this very beautiful nation, working in what was fairly new terrain for me. I was a women's rights lawyer heading the women's rights team at ActionAid International, living in Ghana. ActionAid International is an anti-poverty NGO with offices in more or less 45 countries, including a Sydney office here in Australia. I was working for the Johannesburg headquarters out of their office in Accra. These are some of the photos I took during my time in Ghana.

For an anti-poverty NGO that had largely been focused on rural poverty, we had just turned our minds to the kind of urban poverty that women were experiencing every day living in big cities. Over the years that followed, I was very privileged to meet with countless women who would tell me stories of their experiences of urban poverty. From garment workers in Cambodia living in the export processing zones, through to young women and girls living in Johannesburg in South Africa, where they had these daily experiences of water and electricity cuts, forcing them to use public toilets and public showers, and often negotiate access to them because they had been taken over by groups of men who would bribe those young girls through various means just so that they could use them.

But it was in Brazil that I saw a level of civil society organisation like no other. In the slums or favelas in Rio, in São Paulo, in Heliópolis, I witnessed the outcomes of a neighbourhood community taking back its city. In a lot of these slum communities, women and men really mobilised to activate governments to actually deliver change. In Heliópolis, in São Paulo, the community really took back its public space. They were making demands for the government of Heliópolis to address inadequate public housing, really poor streets, issues around health and education. On my visit to Heliópolis in São Paulo, I actually got to visit the community centre that resulted from those years of advocacy. In that community centre was a small community library and a bunch of computers. But it was really the women of Heliópolis that I think benefited the most from this advocacy. Childcare was introduced as a public service. After one of their most successful night marches, the government of Heliópolis started to prioritise repairing streetlights. These were smart and brave women, and I think they were changing their daily experiences of safety and security.

You might be wondering why I'm talking about this particular story as part of a talk for Tech and Social Justice Week. Well, it became really apparent very quickly to me that a new development was emerging in the way women's groups were claiming their rights and demands for change. You have to remember, this was back in 2010, 2011, and a lot has happened in terms of technology since then. Remember, the iPhone was only released in 2007. Some of you might have heard of the Safe Cities apps that flourished in the months or years that followed back then, 13 years ago. At ActionAid, I headed up a multi-country program on Safe Cities. We piloted this program in Bangladesh, Cambodia, India, South Africa and Brazil. At the same time, the United Nations, through UN Women, were running a similar program. Women from all around the world—in Delhi, Istanbul, London, and even here in Sydney and Melbourne—started to campaign using mobile phone apps. They would mark unsafe streets with crosses, or unlit bus shelters, or places where public transport stopped at a particular time of day, which would obviously affect certain women who needed it the most, particularly women who were shift workers. In a really extraordinary way, technology was offering women a new solution—it was sort of like a virtual protest.

However, the whole response also raised a bigger issue: who would be responsible if the markings on the maps actually alerted abusers to the least lit streets, and therefore the riskiest places where women moved? Very quickly, we realised these women were marking these unsafe places on apps that were public, and potential abusers—who were mostly men who might have wanted to target women—suddenly knew where to go. So this was my first real foray into women's rights and technology.

I've always wanted to be a women's rights lawyer, and I had instilled in me from a very young age by my parents, who were migrants to Australia in the 1970s, this need to be conscious of those around us with less. For me, law—which I think has incredible potential to advance social justice—has always been my natural terrain. Technology, if you know me, you would know is certainly not. So this shift into technology and law has certainly been less planned for me. I imagine I'm not the only feminist lawyer who has found herself at this unusual crossroads between technology and women and law. After all, this language of "femtech", which was really coined to talk about the way in which women's health has been advanced through technology, only really emerged in 2016. And seven years on, I think women's groups are much more cognisant of the potential behind technology for women's health, but also all the challenges and risks that it poses, especially from a privacy perspective.

So fast forward from those early years at ActionAid to 2017, when after a decade living in, I think, pretty much every inhabitable continent in the world, I returned to Sydney and I joined the University of Technology Sydney. I think this move back into academia made so much sense for me. I'm really so well supported by my cohort of socio-legal scholars like Jane, who do brilliant work, and many feminist socio-legal researchers. So I just build on pathways that I think have been walked long before me. Also, given the technology in UTS's name and our commitment to social justice, I think this marriage between tech and social justice in my work has a really natural home, even if my journey hasn't always been an easy one—and I'll talk about that journey a little bit later.

As Jane mentioned, earlier last year, and to my great shock, I won this award as a Woman in AI in the law category. But I think probably in reality, my work is better described as a woman working with AI and using AI for bigger goals. My goal in my work is to help fight gender inequality through the law, and I'm willing to try any means that might work, new and old, that I think legislators and policymakers can take on board to advance gender equality. In that sense, I really want to exploit the incredible potential I see when law and technology come together.

So I want to tell you a little bit about how I came to that point as the creator of the Gender Legislative Index. My starting point was really a social justice one. But hopefully we'll arrive at the point where you can see the potential I see when technology and law come together, but also what I think we're up against.

I consider myself an Australian scholar of gender equality. If pressed to give an elevator pitch, my research studies how to bring a gender lens to legislation, what tools legislators need to make that a reality, what are some of the barriers, and if those barriers can be overcome, what are the benefits for women? If you were to catch a second lift with me, I would tell you that I created the Gender Legislative Index to measure the gender responsiveness of domestic legislation when measured against women's rights standards. It uses human evaluators and machine learning to do that. So imagine now that the lift gets stuck. I'll explain to you why I think tools like the Gender Legislative Index actually matter to a country like Australia.

Some of you might be familiar with the World Economic Forum's Global Gender Gap Index. If you're not, it's definitely worth spending some time having a look. The Global Gender Gap Index made attention-grabbing headlines when they declared that the world is 135 years away from gender parity on a global scale. That figure has more or less shifted from between 100 years and 200 years, but to put it simply, we are far away from where we need to be. The Global Gender Gap Index has been released by the World Economic Forum almost every year since 2006. It ranks countries for their performance on gender equality overall. It has all these sub-ranks on women's economic participation, on women's educational attainment—something Australia actually does really well on—political participation, and then health and well-being.

In comparative terms, Australia fell from a rank of 15th back in 2006, just after Canada, to a rank of 50th in 2021. We had a slight improvement to 43rd in 2022. I will say that in 2023, our rank improved considerably to 26th—that's between Mozambique in 25th and Chile in 27th. I think a big part of why our rank improved so much is that after the election at the federal level, there's been a huge increase in the number of women in politics, which is one of the measures of the index.

This rank in the Global Gender Gap Index is a rank of relative inequality. What this means is that the inequality gap between Australian men and Australian women—because it measures in binaries—is bigger than the inequality gap than all the other countries that rank before us. It comes as no surprise that Australia ranks behind Norway, Finland, Sweden and Iceland, and the kind of countries we assume are gender equal. But Australia's rank also puts us behind Rwanda, the Philippines, New Zealand, Mozambique and a whole host of other countries. What's particularly concerning is that according to the Organisation for Economic Cooperation and Development, Australia is one of the ten richest countries in the world based on GDP per capita. Basically, what that means is we can do better. Yet we perform particularly badly when compared to countries with the same levels of economic development as our own.

As someone who spent a long time living overseas, to me it's really important to acknowledge that gender equality is obviously not a third world problem—it's not a global South problem—it's a global problem. Although we need to recognise that the way women experience inequality is very different in a country like Australia than it is elsewhere. But I think it's so evident how important gender inequality is and how prevalent it is in a country like Australia after everything we've seen in these last few years. Everything from Me Too, but also when we acknowledge which women in Australia are being left behind. If we think about the rapid rise in homelessness among older women in Australia, or the inequality gaps experienced by First Nations women.

So I effectively created the Gender Legislative Index as a tool to potentially do something about this. I started my research on women's rights. I got to this point where I thought I want to create a tool to measure whether law can play a more fundamental role. I think law hasn't reached its optimum potential in this fight for gender equality. So I started evaluating legislations, I created scores, I had pieces of legislation and pieces of paper, and I was mocking up things on Excel. I got to the stage where I thought, I think this could work—this tool to measure legislation and compare provisions in laws. But my skills as a women's rights lawyer sort of ended there, and I needed collaborators. I needed collaborators to take this into something that had scale and that could actually send a message to people in a way that might lead to change.

So I looked for collaborators and ended up working with the UTS Connected Intelligence Centre, which is a team of data scientists who do amazing data analysis and data visualisation work. They're based just on the other side of Broadway, and with Rapido Social, who sit in the Faculty of Engineering and IT, actually in our building too. Rapido do all sorts of incredible things—they build bionic limbs, but they also do a lot of software engineering and have helped our very own Anti-Slavery Australia, but also have collaborated with me on the Gender Legislative Index. I owe many thanks to my collaborators for believing in my vision for a tool to demonstrate better whether law is playing its part in correcting inequality and making women's lives better.

So the GLI, or the Gender Legislative Index, is an interface to measure how well domestic laws actually take into account the differences between men and women, and the needs of specific groups of women, especially those women who are being left behind. It's powered by a mix of human evaluators and machine learning. I'd like to take you through step by step how the Gender Legislative Index works, what sort of evaluations have come out of it, and then what impact it's had. Then I'm going to turn in the last part of my presentation to this bigger question of gender and AI.

Effectively, we have a team of evaluators, and they evaluate the actual provisions of pieces of legislation. They do that by asking these seven questions on the screen here. They ask questions like: does this law actually guarantee access to services? Does it guarantee access to information? Does it protect women from non-coerced decision making and promote equality? Those seven questions have been derived from a study that I did of the Convention on the Elimination of All Forms of Discrimination Against Women.

When evaluators answer those seven questions, they place their answer on a scale from gender regressive right through to gender responsive. Obviously, whether a service is available in a law or whether a law promotes equality is going to differ depending on whether the law is about reproductive health, tax, or mining. The Gender Legislative Index has built in, thanks to a team of researchers who worked with me, a set of benchmarks so that the evaluators can make their evaluation based on good practice in a particular area of law.

Once the evaluators place their answer on a scale from regressive to responsive, they give an overall assessment of a law, deciding whether it shows a complete disregard for international standards right through to meeting international standards. Effectively, the evaluation team has shown that laws in Australia, in the context of workplace health and safety, don't actually count gender-based violence in the way they treat dangers or risks on the factory floor. Our Work Health and Safety Act tends to look at lack of safety in terms of accidents or spillages and workplace hazards, and not the kind of harassment that women might be experiencing in those workplaces.

Evaluators have shown that Australia's paid parental leave laws set women as the primary carer and promote this suggestion that women, as opposed to equal caring between parents, are responsible for children. That's really been the case until this year when Australia's paid parental leave scheme has changed—it's been reformed. Evaluators have pointed out that Australian tax laws that have a dependent spouse tax offset discriminate against non-married women and often encourage women in heterosexual relationships to work fewer hours or part-time, creating greater financial insecurity among women.

As is so often the case, as you can see, the Gender Legislative Index has a lot of data in it—there's laws from Australia that have been evaluated, but also Indonesia, the Philippines and Sri Lanka. I often had people say to me, "Well, which country scores the best? Which laws are the best?" People wanted a bigger headline story. This is where my work with data scientists really began.

I started working with a data scientist and very quickly realised that lawyers and data scientists speak very different languages. It sounds a bit like a bad joke—a lawyer and a data scientist met in a bar. We didn't actually meet in a bar, but I have to say I was very privileged to work with a data scientist who genuinely wanted to understand our science of the law. He would sit and ask me questions: "What makes this law regressive for women? How does that work in practice? Is gender neutral a good thing or a bad thing? Is it just a cover for a gender blind law?" We had many, many conversations of that kind and we reached two solutions.

The first solution was to use heat mapping. Heat mapping allowed us to show when there was agreement or disagreement among evaluators, because lawyers often disagree on things. I didn't want to hide that data in the Gender Legislative Index. The heat mapping tool shows how much agreement there is among evaluators about a particular score. This is what the GLI looked like as a first mock-up—it was pretty simple, all coded in blue. You could see levels of agreement or disagreement among evaluators. In the bottom half of the screen, this is how the Gender Legislative Index looks now, because we used user experience expertise to change some of the colouring and make it just look a bit more accessible to a public audience.

What really sets the Gender Legislative Index apart in terms of indices is it has an algorithm built in. The algorithm allowed me to take all of this data and give a final score. We can have three evaluators, we can have ten evaluators—but what does it all mean? How do you pull it all together? Basically, the Gender Legislative Index's algorithm operates as a series of ordered decisions. This can be quite complicated to understand, but if you think about it as a decision tree, it becomes really simple. A particular evaluator might give three gender blinds and two regressives and one neutral and flow through to a final score—in this case, not meeting international standards. The way in which the algorithm operates, it also operates as a decision tree and gives a final score for each law.

What's a bit more special about using an algorithm is that the algorithm is trained on a subset of data in the Gender Legislative Index, but the algorithm doesn't know which law is being evaluated in that particular case or whether the data reflects a particular country or a particular leader was in power at the same time. The data on which the algorithm is trained treats all the laws the same. Part of the reason for this is to reduce some of the subjectivity that we bring as human evaluators to a piece of legislation, because we tend to evaluate certain countries better because we assume they have better laws. I've seen how evaluators tend to evaluate laws on gender-based violence more harshly than a tax law, maybe because we have more experience in that field. So I hope that the algorithm in the GLI brings a bit more integrity to the final score.

With all of that tech in the back of our minds, I guess the bigger question is, well, what has it led to? As Jane very kindly pointed out in the introduction, it's had an impact on the way laws are made, particularly in the state of Tasmania. A couple of years ago, an independent member of parliament, Ruth Forrest, from the Tasmanian Legislative Council, heard about the Gender Legislative Index. She decided to issue a motion in the Tasmanian Legislative Council for Tasmania to establish a joint sessional Gender and Equality Audit Committee. She cited my work, saying that Australian laws shouldn't be written as if men and women are the same and that gender differences don't count. That motion was successful.

Now Tasmania's Gender and Equality Audit Committee is the only state with an audit committee to look at legislation in a parliament. The ACT has one on gender and economic equality, but there's no other standalone committee focused on gender at a state or federal level. That's a photo of Ruth and I on the right after she invited me to speak to the Commonwealth Women Parliamentarians—that's women parliamentarians from all state levels and the federal level—to talk about the GLI.

About a month ago, I was very privileged to be invited by the federal Department of Prime Minister and Cabinet's Office for Women to give a talk at the Asia-Pacific Economic Cooperation Conference, the summit in Seattle. The APEC summit is a meeting of around 21 economies, including Australia, and Australia co-chairs a special working group with the US on women and the economy. I went over to Seattle and talked about the Gender Legislative Index and gender budgeting. The chair of that committee has now put forward the idea of gender-responsive lawmaking to the Women's Economic Equality Taskforce that's reporting to our Prime Minister. I've been told it should be appearing in the new Gender Equality National Strategy, and that there's a call for a federal audit of all Australian legislation from a gender perspective and hopefully a federal audit committee. So it's a bit of a watch this space, but I have to personally say it's been really rewarding to see the research come off the page and have this impact on lawmaking in Australia.

I could talk quite endlessly about the Gender Legislative Index, but I wanted to use the rest of my time to talk a little bit more generally about AI and gender. As you can see, I sort of stumbled into machine learning a little bit by accident. I remember when I first started working with Mike Prussy, the data scientist at KIC, this was in 2018 to 2020—so this was pre-ChatGPT. At that time, I don't think many people realised just how present AI already was in our lives, let alone could actually explain how it worked. I think people would have considered it science fiction.

So I found myself at this point where I understood or began to understand how machine learning worked, and I was calling for a gender lens to legislation. I thought, well, I've got to bring this together. I started looking at what governments like Australia and a whole host of other governments who are trying to legislate on AI in a sort of race to regulate could actually do with those laws to bring into legislation a gender perspective to address some of the gender harms of AI. I just want to talk a little bit about some of those harms and then finish up by talking about what all of us can do about them.

I think there's a lot of, I would dare say, scaremongering and fear online, especially around the risks of AI to gender. Some of those risks are very real, and I think it's important for us to acknowledge that so many of those gender-based harms of AI actually started long before AI, but AI tends to replicate inequality that exists in our society for women and for gender diverse people. There are really two harms that people talk about a lot: allocative harms and representative harms, and one that I don't think gets enough attention, which is knowledge inequality harms.

Some of the allocative harms of AI arise from a gender perspective when decisions are made about how to allocate goods. The one that gets probably the most attention is when AI-driven technologies are used by banks to determine the creditworthiness of applicants. If those AI-driven technologies are trained on years of data that was denying certain women bank loans, then that inequality and that discrimination risk being perpetuated by the dataset, meaning that certain women, such as women sole heads of households or women from minority communities, might continue to get denied a bank loan.

Another really common example is the use of AI-driven recruitment tools, which flourished after COVID because everyone started recruiting online and they became really popular. A lot of these AI-driven technologies are also trained on years of CVs where men dominate and women don't, and they might actually filter out women's CVs altogether. Some of the early technologies were found to filter out of consideration any CV that mentioned "women"—even if that was "women chess club champion" or "graduate from a women's college". They're the kind of allocative harms that affect women's lives.

There's a whole bunch of representative harms from AI that reinforce women's and gender diverse people's secondary status through stereotyping or underrepresentation or denigration. There are a whole lot of examples that fall into this category, one of which is deepfakes. These fake videos use images or videos and present them as videos that look very real. It's deepfake pornographic videos where women's footage has been used in a non-consensual way that are probably the most harmful ones in this category. There's a whole lot of other ones that suggest it's women who are nurses and not firefighters, or black men are more likely to perpetrate crimes than white men. So there's also a huge racial bias that's part of AI's makeup as well.

Another example in this field: female-voiced assistant technologies. I have a Google Home and she speaks with a female voice. Some people might think, is it that big a deal that Siri and Cortana and Google all have female voices? A whole group of scholars have started writing about the ways in which those female-voiced assistant technologies perpetuate stereotypes about women—that we're just here to serve. What's really interesting is when technologists decide to use a male voice—IBM's Watson uses a male voice to train doctors on how to treat cancer patients. There's a lot happening in that space, and I think what's even more concerning is that some of these female-voiced assistant technologies and female-named bots are now being the subject of harassment. The kind of harassment women experience in everyday life is now being projected to female-voiced assistant technologies and is really normalising that in everyday life.

The third harm I mentioned was knowledge inequality harms, and I'm going to come back to that one a little bit later. Having spent so long talking about the Gender Legislative Index as a tool that uses machine learning for social justice and technology for social justice, I just wanted to balance it out by talking about some of these technology-driven harms for women.

I felt particularly alarmed last year when my very first job in human rights, which was at the Center for Reproductive Rights in New York, intersected with my more recent work on AI. Some of you might have seen the news that went around when police in the US were using AI-driven technologies and menstrual apps, which women were using online, to actually track which women were accessing abortion services in the US, particularly in states where access to safe and legal abortion is so restricted. When these women's groups were suddenly saying "delete your menstrual apps and delete all your data", it's unsurprising that people are so fearful of the harms that AI can produce for women and gender diverse people.

Again, I also think that the goal at the end of the day has to be in seeing that good in AI and balancing it with the harm. I do want to offer three more examples, because I shared the Gender Legislative Index as my example of using AI for social justice, but there are so many more. There's so much potential here if we can see it and understand how to exploit the technology for gender justice.

In Spain, there's an algorithm being developed called VioGén—violencia de género, so gender-based violence. The algor