“The choices made today regarding AI will shape whether this critical technology amplifies new and existing inequalities at scale — or contributes decisively to a more inclusive future.”

On Thursday 15 October 2026, the Victorian Women’s Trust hosted a conversation on one of the most urgent questions of our time: what does the rapid rise of artificial intelligence mean for women, girls and gender-diverse people?

UTS Chair in Gender and the Law, Professor Ramona Vijeyarasa, joined the Victorian Women’s Trust’s inaugural Feminist Researcher in Residence, Dr Elise Stephenson, to explore the gendered harms of AI, why Australia’s approach remains largely gender blind, and what can be done to address these challenges.

Gender and AI Webinar Opening Screen

Gender and AI: Where's the Harm? Webinar Recording

(00:55:13)

Gender and AI: Where's the Harm? Webinar Recording transcript

Good afternoon everyone and welcome. It is an absolute pleasure to join you on behalf of the Victorian Women's Trust.
0:13
13 seconds
My name is Ramona Ramona Bijratha and I'm a professor in the faculty of law at the University of Technology Sydney
0:20
20 seconds
where I hold the inaugural chair in gender and the law. The question before us for today's conversation is gender
0:27
27 seconds
and I AI where's the harm? And I think we all know it's an urgent question.
0:32
32 seconds
We've had so many people interested in today's conversation and I think it's a reflection of that urgency. AI is being
0:39
39 seconds
rapidly embedded in the institutions and technologies and everyday decisions that are shaping our lives. It's influencing
0:47
47 seconds
what we see, the information we receive, the opportunities available to us, how we are assessed, and how decisions are
0:55
55 seconds
being made about us. And I think even more so, this is happening at a moment of profound backlash against women's
1:03
1 minute, 3 seconds
rights and gender equality, diversity, and inclusion. AI is not entering an equal world. is entering societies where
1:11
1 minute, 11 seconds
we know that the institutions and the labor markets and the legal systems are already distributing power and opportunities in really unequal ways.
1:18
1 minute, 18 seconds
And so unless we intervene deliberately as we are doing today, AI risk reproducing those inequalities,
1:25
1 minute, 25 seconds
accelerating them and making them more difficult to see and challenge. So a great deal of important work is about
1:33
1 minute, 33 seconds
making those gendered harms of AI visible which is the topic of today's discussion. and this excellent research snapshot um produced by the Victorian
1:42
1 minute, 42 seconds
Women's Trust and authored by Elise who I'm in conversation with today. My own work tends to begin with the question that follows which is once we see the harm, what should the law do about it?
1:53
1 minute, 53 seconds
And it's not simply a question of whether we need more regulation. It's a question of what kind of regulation we need. The law can apply the same rules
2:01
2 minutes, 1 second
to everyone and still fail to recognize who experiences harms most acutely whose
2:08
2 minutes, 8 seconds
interests remain invisible and who can realistically actually obtain a remedy.
2:14
2 minutes, 14 seconds
So I'm hoping today's conversation brings together two closely connected tasks. The first is to identify and understand the gendered harms of AI. The
2:23
2 minutes, 23 seconds
second is to consider how do we translate that understanding into accountability. Um, but before I begin, I would like to acknowledge country.
2:31
2 minutes, 31 seconds
I've already seen we have guests joining from all around the world. That's incredibly exciting. Um, especially the time zone that we're in over here on the
2:38
2 minutes, 38 seconds
east coast of Australia. I'm joining from the University of Technology Sydney Faculty of Law. And the UTS Faculty of
2:45
2 minutes, 45 seconds
Law is situated on the unseated lands of the Gatagle people of the Euro Nation.
2:50
2 minutes, 50 seconds
And I pay my respects to elders past and present. And at UTS, we acknowledge that the lands have always been places of law
2:57
2 minutes, 57 seconds
and continue to exist in plural legal worlds. And for those who are less familiar with Australia's indigenous culture, I think it's important to
3:04
3 minutes, 4 seconds
reflect that today is actually the 15th of October, the International Day of Rural Women. And it's a day that
3:11
3 minutes, 11 seconds
recognizes the the vital contributions especially indigenous rural women make to things such as food security and rural development and biodiversity. And
3:20
3 minutes, 20 seconds
I think this is actually really relevant to our conversation today. And I'm sure Elise agrees with this sentiment that debates about gender and AI can't just
3:28
3 minutes, 28 seconds
be about who has access to the tech. For rural and indigenous women, digital systems so often draw upon their knowledge of land and country and
3:37
3 minutes, 37 seconds
communities, but without always adequately recognizing or getting their consent. and a genuinely rightsbased approach in this digital world is not
3:46
3 minutes, 46 seconds
just about digital access but indigenous sovereignty over knowledge and data and self-determination.
3:53
3 minutes, 53 seconds
So with that in mind, because I know this is very relevant to Alisa's other work, it is really my pleasure to introduce my friend and colleague, the
4:00
4 minutes
wonderful Elise Stevenson. So I don't think Elise needs much of an introduction but she's the director the deputy director of the global institute
4:07
4 minutes, 7 seconds
for women's leadership at the Australian national university and an Australian research council decreello her research
4:14
4 minutes, 14 seconds
examines gender inequality at what I think are three fundamental emerging policy frontiers so AI space and climate
4:23
4 minutes, 23 seconds
and Elise just released her publication space xxx gender power and the new space race I urge you all to go out and get a copy and I had the priv privilege of
4:32
4 minutes, 32 seconds
reading being an early copy of Alisa's book. And I think SpaceX really reflects what makes Alisa's scholarship so
4:39
4 minutes, 39 seconds
compelling. This capacity to bring together gender, power, and inequality to fields that are so often treated as
4:46
4 minutes, 46 seconds
purely technical or neutral or somehow beyond politics. Elisa is also the inaugural fellow um or resident, I
4:54
4 minutes, 54 seconds
should say, the feminist residence researcher in residence at the Victorian Women's Trust. And this residency is not
5:02
5 minutes, 2 seconds
concerned only with documenting inequality. I think as would be expected of the Victorian Women's Trust, the residency is designed to turn rigorous
5:10
5 minutes, 10 seconds
feminist research into advocacy and practical change. And so the research snapshot we'll be hearing more about today is the foundation for a white paper that will be released next year.
5:21
5 minutes, 21 seconds
So we're going to discuss the research snapshot, gender and AI, where's the harm? And really it maps these 20 interconnected gendered harms and shows
5:30
5 minutes, 30 seconds
why they can't be treated as a collection of isolated technical failures. So Elise, a warm welcome and let me begin with the question that I
5:38
5 minutes, 38 seconds
think is at the heart of this snapshot, which is what were you trying to uncover through this project and what did you find, you know, that surprised you the most when you did this research?
5:47
5 minutes, 47 seconds
Thanks so much, Ramona, and thank you to everyone who's joined us and also the Victorian Women's Trust for really backing this research in the first um
5:54
5 minutes, 54 seconds
place. I'm joining from Nano country um in Canra, Australia. For me, this research I mean you mentioned it a
6:02
6 minutes, 2 seconds
little bit in terms of my broader research agenda which has really you know it's kind of picked on what could
6:08
6 minutes, 8 seconds
be seen as unrelated um domains. So space AI climate change but actually they're deeply related and they go to
6:17
6 minutes, 17 seconds
the heart of a question that I've pondered for some time which is the way I see it. We are in a critical juncture
6:24
6 minutes, 24 seconds
where we've got new institutions emerging, new rules, new policies, new legislation, new departments. um we have
6:31
6 minutes, 31 seconds
a flood of resources and funding that are going into areas like artificial intelligence and some of the techni technological advancements and in kind
6:41
6 minutes, 41 seconds
of theoretical terms and you know I'm a political scientist and an international relations scholar and what we really study is institutions and seeing kind of
6:50
6 minutes, 50 seconds
these times where you've got new institutions or where you've got a flood of funding and kind of new resourcing and new actors as critical junctures
6:57
6 minutes, 57 seconds
where as the theory goes new institutions created in more gender equal times should be more gender equal.
7:06
7 minutes, 6 seconds
And I really wanted to critique that and question that and and understand at the heart of it um are we always doomed to
7:13
7 minutes, 13 seconds
repeat inequalities or can we indeed do things better? Can we learn from past mistakes and injustices and create better pathways forwards in the future?
7:22
7 minutes, 22 seconds
So, this Victorian Women's Trust uh residency has been a real pleasure and an opportunity to kind of stretch the work that I've been doing previously
7:30
7 minutes, 30 seconds
anyway and look specifically at um mapping at a systems level the gendered harms of AI because you know you might
7:39
7 minutes, 39 seconds
hear a little bit about data centers over here, deep fakes over there, maybe biased LLMs. Um you often hear about the
7:46
7 minutes, 46 seconds
the challenges around gender and AI or AI more broadly in silos. But what I really felt was missing was this connected picture. What are the
7:55
7 minutes, 55 seconds
structural ways in which AI is kind of enhancing inequality or kind of embedding inequality as well as of
8:03
8 minutes, 3 seconds
course how can it be disrupted. Um and and I think that that's been kind of something that's underpinned the research is that overarching picture but
8:12
8 minutes, 12 seconds
also the search for feminist leverage points within this system. we don't have to do it the way that we always have
8:20
8 minutes, 20 seconds
done it and uh by taking a explicitly feminist AI approach. It might include um doing things like um challenging the
8:29
8 minutes, 29 seconds
neutrality of technology you know really revealing that technology isn't neutral.
8:33
8 minutes, 33 seconds
It is a reflection of kind of who makes it but also the biases that are inbuilt into it and kind of society more generally. Um as well as not all forms
8:42
8 minutes, 42 seconds
of technology are inevitable. you know, we don't see um uh you know, I don't know, let's think nuclear or some of
8:49
8 minutes, 49 seconds
these other areas. There are really heavy guard rails around some forms of technology for good reason because not
8:56
8 minutes, 56 seconds
only um may they have kind of uh significant kind of severity or significance now, but also for a long
9:03
9 minutes, 3 seconds
term into the future. So for me it's really about um mapping those harms at a structural level, identifying the feminist leverage points and really
9:11
9 minutes, 11 seconds
thinking in an integrated um intergenerational approach to what does a feminist AI approach look like for our world.
9:19
9 minutes, 19 seconds
Well, I want to dig into those mappings and and ask, you know, why why was it time now to to bring it all together?
9:25
9 minutes, 25 seconds
But first, I would love to hear more about those feminist leverage points because I think that will get everyone excited and this idea of technology not
9:32
9 minutes, 32 seconds
being genderneutral. I see that written all the time now. People are recognizing that tech is not gender neutral. But then therefore what next?
9:43
9 minutes, 43 seconds
Yeah, absolutely. And so this is really a big big part of the the kind of opportunity or discussion. So um with
9:50
9 minutes, 50 seconds
this particular project uh given there was kind of siloed pockets of research here there and everywhere um by trying to bring kind of these 20 gendered harms
9:59
9 minutes, 59 seconds
together and I recognize that's not exhaustive. I'm actually hoping that people take this work and they add to it, maybe they tweak it. Um, you know,
10:07
10 minutes, 7 seconds
that this is maybe a starting point more than anything else. Um, but I think that what it allows us to do is see the problem clearly. I mean, as you said at
10:15
10 minutes, 15 seconds
the start, uh, in order to find leverage points, you have to know kind of root cause issues and understand the nature
10:23
10 minutes, 23 seconds
of the problem. One of the things that I've encountered in a lot of my research is that we might have gender expertise over here. We might have artificial
10:30
10 minutes, 30 seconds
intelligence expertise over here and we might have policymaking expertise over here. Very rarely do you find people at this intersection of expertise and
10:38
10 minutes, 38 seconds
that's kind of for a reason, right? So although artificial intelligence has been around for decades now, um we've seen in the recent years a real ramping
10:47
10 minutes, 47 seconds
up of its use and um kind of mainstreaming of uh for instance large language models and chat bots as well as
10:54
10 minutes, 54 seconds
integration into um you know I don't know physical products and you know fridges and cars, TVs, you know all of the rest um which make it much more
11:03
11 minutes, 3 seconds
prevalent and kind of change our interaction. Um but you know even so um I think there's an emergency emerging
11:12
11 minutes, 12 seconds
evidence base and that is something that we've tried to capture with this snapshot is that some of the harms um
11:19
11 minutes, 19 seconds
are absolutely already here and have a strong and substantive evidence base which we we can you know go ahead and
11:26
11 minutes, 26 seconds
make policy on immediately. Some of the harms identified are a little bit more theoretical. they could potentially happen. Um, even if they haven't
11:35
11 minutes, 35 seconds
happened now, they warrant inclusion because of the devastating nature of the harms. And so I've tried to map kind of
11:42
11 minutes, 42 seconds
across, you know, the strength of evidence, across the severity of or significance of the harm, but then also the durability of the harm. I mean, I I
11:50
11 minutes, 50 seconds
used the chat um the example uh last week when we were launching um this uh publication. Um I used the example of um
11:58
11 minutes, 58 seconds
a chatbot that misgenders you. Um, so that harm might actually be relatively easy to fix by training a chatbot not to
12:08
12 minutes, 8 seconds
do that, right? So the durability of harm isn't as persistent as for instance um artificial intelligence or LLM models
12:15
12 minutes, 15 seconds
that have been built on you know um biased data sets and that continue to reinforce certain norms that you know
12:24
12 minutes, 24 seconds
forever exclude women or you know first nations voices or whatever it is that might have a much longer term impact that takes a lot longer to change. So
12:33
12 minutes, 33 seconds
across those hearts, I think that the leverage points look different. And what I'm hoping comes from this snapshot next is firstly, we're going to run some
12:42
12 minutes, 42 seconds
workshops and get some people in a room to try and map out what are some of the policy options. But also I hope that people wherever you might be in the
12:49
12 minutes, 49 seconds
world um have a read through um this research and say hey okay for my community I think we could do this to
12:56
12 minutes, 56 seconds
start to uh address this point specifically say in schools or for those who work in some industry um you know
13:04
13 minutes, 4 seconds
what is it that we can change in our um I don't know development practices or maybe in our hiring practices or what have you. So in some ways um I can help
13:13
13 minutes, 13 seconds
to co collate some of those leverage points. I'm also hoping that the community takes this upon themselves to identify some feminist leverage points too.
13:21
13 minutes, 21 seconds
Excellent. And I think it's a really empowering opportunity for many of our listeners. In the examples you gave, you know, you've just hinted at some of the diversity of the issues we have at hand.
13:31
13 minutes, 31 seconds
We tend to talk about AI as an umbrella term and now we're talking about gender and AI or the gendered harms of AI.
13:38
13 minutes, 38 seconds
There's a lot of within that umbrella.
13:40
13 minutes, 40 seconds
And so you've talked about chat bots that misgender. You've talked about LLMs that rely on biased data and maybe making decisions about e education
13:49
13 minutes, 49 seconds
access or finance or whose CV makes it to the short list pile. Um what has been the risk of us looking at these in a
13:57
13 minutes, 57 seconds
really siloized way and what's the value of bringing it all together to talk collectively about gendered harms?
14:05
14 minutes, 5 seconds
Yeah. Well, I think looking at it in a siloed way is what we've seen to date in Australia as well as many other global um uh kind of counterparts. You'll know
14:14
14 minutes, 14 seconds
this better than I do, but what we often see is maybe one department specifically deals with just the copyright issues or
14:21
14 minutes, 21 seconds
maybe I don't know another regulator just deals with um uh you know image based abuse and deep fakes. um that's
14:30
14 minutes, 30 seconds
fine, but in lacking kind of a more holistic approach to the gendered harms of AI, um it it's also a little bit
14:38
14 minutes, 38 seconds
reflective of our kind of governance or policy approach, which is well, we've got gaps. There might be things that are
14:45
14 minutes, 45 seconds
missed and and particularly for me, I think um really this mirrors inequality broader in society. So, we're not just
14:53
14 minutes, 53 seconds
seeing um I suppose what I don't want is just a siloed approach that treats deep fakes as the issue and it starts here
15:02
15 minutes, 2 seconds
and it ends there and kind of um I don't know that by by looking at it so narrowly you might miss all of the
15:09
15 minutes, 9 seconds
preceding factors that enable this and all of the kind of um other ways in which harms are kind of embedded across the ecosystem. So for me, I would love
15:18
15 minutes, 18 seconds
it if this isn't just the responsibility of um tech departments and you know certain regulators, but that there is a more mainstreamed approach and that
15:27
15 minutes, 27 seconds
society has a way of looking at AI and understanding it for both its opportunities and its harms. Um I think
15:34
15 minutes, 34 seconds
a lot of the research that I've done so far in interviews and focus groups has really kind of revealed a few things. I mean, you know, some things that, um, a
15:42
15 minutes, 42 seconds
lot of women kept, uh, telling me in interviews, uh, which absolutely needs further research to actually establish prevalence, but, you know, I thought it
15:50
15 minutes, 50 seconds
was interesting, um, was, you know, women were saying, "Hey, I I'm taking on the AI mental load here." When it comes to actually thinking about the ethics of
15:59
15 minutes, 59 seconds
AI, I seem to be the only one thinking about it. Um, I don't see industry thinking about it. Sometimes I don't see my men male colleagues thinking about it as much, and I'm worried about that.
16:08
16 minutes, 8 seconds
Now, that's a replication of what we see elsewhere in society when we talk about mental load, uh, gendered mental load, um, unpaid caring burden, you know, all
16:16
16 minutes, 16 seconds
of these other things. So, what I'm hoping to show is that this is not, um, removed from what we see in society more broadly. It's an extension. The way we
16:25
16 minutes, 25 seconds
theorize it at the Global Institute for Women's Leadership is that technology is a is a tool that can either um change
16:34
16 minutes, 34 seconds
the nature or prevalence of harm. And I think that that's kind of probably at the key point there is that prevalence
16:42
16 minutes, 42 seconds
and nature of harm and that there are things that we can do because technology is not innate um natural something that
16:49
16 minutes, 49 seconds
came into the world you know and there it is. It's something we created and we co-create at every juncture. Um, so by
16:56
16 minutes, 56 seconds
seeing kind of interconnected harms, we also can see how interconnected little tweaks here, here, here, here, and here
17:03
17 minutes, 3 seconds
can help to create a more just, equal, and sustainable future going forwards.
17:08
17 minutes, 8 seconds
Thanks, Elise. I find that example about the mental load really interesting and very reflective of what the literature and scholarship says in the space
17:16
17 minutes, 16 seconds
already. In response to the fact that women only make up around 20 22% of PhD graduates in AI and computing science in
17:25
17 minutes, 25 seconds
the US, Susan Levy, who's a who's a technologist, she's a computing scientist, has made a strong push for more women at the table. And many say,
17:33
17 minutes, 33 seconds
well, that's not necessarily going to change the problem. Just having more women, and we know just having more women at a table won't in and of itself um be the answer to to these problems.
17:42
17 minutes, 42 seconds
But she points out that the vast majority of scholars talking about the gendered harms of AI are in fact women and gender diverse people. Partly
17:50
17 minutes, 50 seconds
because it's lived experience and partly because we're bearing that responsibility to bring these issues to the table. And she therefore says, well, if we're the ones conscious of it
17:58
17 minutes, 58 seconds
because we have this lived experience, we need to be at the table to raise it.
18:01
18 minutes, 1 second
So I think there is evidence to back that women and gender diverse people really are carrying the mental load of
18:07
18 minutes, 7 seconds
these issues. You talked before about um the nature of the harms themselves and in the report it's described as as a
18:16
18 minutes, 16 seconds
cascading system which I think um is something to to alarm us and to bring us to this discussion and and to take it really seriously. I also like that you
18:25
18 minutes, 25 seconds
just talked about creating and and co that that that AI and the technologies is something we're creating and co-creating because I think it also
18:32
18 minutes, 32 seconds
raises the reality that this space is changing so fast. So, companion chat bots are on your scale um in terms of
18:40
18 minutes, 40 seconds
where their their severity and and and significance. And I and I think it's a really interesting example. Someone in my team has done studies on um companion
18:48
18 minutes, 48 seconds
chat bots that enable simulation of of rape and sexual violence and situations tantamount to slavery even if they
18:56
18 minutes, 56 seconds
weren't designed to do that and for those that were designed to do that. and and I think it's an alarming but real example of just how quickly these technologies even in the space of one
19:05
19 minutes, 5 seconds
year have leared to do things that we didn't anticipate. So can you tell the audience a little bit about your cascading system or classification and
19:13
19 minutes, 13 seconds
and how we're meant to understand the severity and the significance of these gendered harms?
19:19
19 minutes, 19 seconds
Yeah, absolutely. Um, and I think that this is something that, you know, in the snapshot you can have a little bit more um of a look into it because one thing I
19:28
19 minutes, 28 seconds
didn't want to create with this snapshot was a hierarchy which says this harm is definitely lower severity or significance than this harm which is
19:36
19 minutes, 36 seconds
definitely higher. And so in in trying to conceptualize this, you know, I've drawn a lot of evidence bases and various theorists, but um thinking about
19:45
19 minutes, 45 seconds
those things that may be on the lower end of the spectrum, maybe those that affect a lower number of people or have a lower intensity of suffering or
19:54
19 minutes, 54 seconds
disadvantage, which again is pretty subjective. So again, we're we're talking really broad categories. Maybe it has a lower impact on rights or autonomy, freedom, safety, or
20:02
20 minutes, 2 seconds
opportunity. or maybe it has a lower impact on reinforcing structural inequalities. Those on the higher end of course would be higher of all of those
20:10
20 minutes, 10 seconds
things that I just mentioned. Um and so I suppose the big caveat is well I think that this really does depend on how
20:18
20 minutes, 18 seconds
technology is used as well as the individual and acknowledging intersectionality and some groups may find that some harms actually are far
20:24
20 minutes, 24 seconds
more significant or severe and have a gr far greater impact on rights and and all of the rest. So, uh, really broad
20:32
20 minutes, 32 seconds
classification which I'm looking forward to others building on. Um, but in terms of that cascading nature of harms, I mean, I think you can, you know, we've
20:41
20 minutes, 41 seconds
talked about a few of these sorts of things already. So, low representation of women in the sector and in AI in Australia, it's about 30%, that
20:49
20 minutes, 49 seconds
decreases quite rapidly once you get to the senior executive roles as well as um, kind of lead uh, professorships and academics who are working in this space.
20:59
20 minutes, 59 seconds
um concerns around ethics, they these sorts of factors, biased LLMs, we see um
21:06
21 minutes, 6 seconds
and again I'm not drawing direct causal links, but I'm absolutely showing what we know. So there's all these things happening. We're also seeing um women
21:14
21 minutes, 14 seconds
opting out of using AI. We're also seeing that because they're not opting into AI, they're not being able to train some of these systems so that some of
21:23
21 minutes, 23 seconds
these systems can learn from maybe more feminist or maybe more ethical approaches. um we see given kind of the
21:30
21 minutes, 30 seconds
the biased nature of these systems, we're seeing things like um uh very uh disinformation, misinformation, biased
21:38
21 minutes, 38 seconds
um advice given and this is on a whole host of um issues. So for instance, one study in Australia found that whilst men
21:46
21 minutes, 46 seconds
and women uh you know use AI similarly when it comes to medical and health advice, women tend to ask um questions
21:54
21 minutes, 54 seconds
of AI that are more um you know that would actually require more um specific expertise or like maybe a more severe
22:03
22 minutes, 3 seconds
sort of health um issue. Um now that should concern us. Um but also when we're thinking about kind of some of
22:10
22 minutes, 10 seconds
those bias systems and you know the how leaders are represented and there's been great studies again I've got in the snapshot um you know women being more
22:18
22 minutes, 18 seconds
often uh kind of pictured as you know less likely to be leaders or you know more demure submissive um so on uh
22:26
22 minutes, 26 seconds
depending on where you're based in the world maybe the job occupations that women are stereotypically um kind of pictured as when it comes image
22:34
22 minutes, 34 seconds
generators uh you know they're not included or they're included as sex workers and cleaners and all of these other sorts of things which are very stereotypical
22:42
22 minutes, 42 seconds
um or sexualized in in cases. Um and so and then also you look at deep fakes,
22:50
22 minutes, 50 seconds
right? You know, you kind of see that these this picture is building. um you know the prevalence of deep fake pornography um really uh basically it's
22:59
22 minutes, 59 seconds
around 97% of all deep fakes are pornographic deep fakes and most of that so about 99% is of women and girls I
23:07
23 minutes, 7 seconds
mean this is horrendous then you link that into political um representation and you see the targeting of leaders so
23:14
23 minutes, 14 seconds
even just in the last week um a huge number of European uh women politicians have been targeted by an onslaught of of
23:23
23 minutes, 23 seconds
really a a targeted campaign of AI deep fake pornographic images. Now, this is horrendous. And so then you see a a
23:31
23 minutes, 31 seconds
chilling effect and we've done a lot of research um at the global institute for women's leadership working with women uh who are either politicians or they're
23:39
23 minutes, 39 seconds
aspiring to be politicians and also working with young people uh gender diverse first nations um LGBTIQ plus and
23:47
23 minutes, 47 seconds
the kind of one of the top reasons why they opt out of a political career or taking more political action um is
23:56
23 minutes, 56 seconds
basically lack of safety and online safety um kind of a fear fear of being targeted, of being doxed, of um you know, having deep fakes made of them,
24:04
24 minutes, 4 seconds
all of the rest. And it's not just a fear that's come out of nowhere. It's a really common um practice that we see in the world. And so this is kind of what I
24:13
24 minutes, 13 seconds
mean by this cascading thing. Um the other interesting thing uh that really came out through the research was well,
24:20
24 minutes, 20 seconds
what's it mean when we rely on AI so heavily for kind of social interactions?
24:25
24 minutes, 25 seconds
So whether that's on Bumble or Grinder or or Tinder for um you know how have you know help me be I don't know answer this response to whomever I'm quoting.
24:37
24 minutes, 37 seconds
You can tell that I'm married and haven't dated in a really long time. I'm so sorry. Um [laughter] um but you know like this socially
24:45
24 minutes, 45 seconds
mediated AI reliance where you have to rely on AI or maybe you don't feel comfortable. Um like what does that do to human relationships over time? and
24:53
24 minutes, 53 seconds
some of the youth advocates I talked to who see this rising prevalence of AI generated um companions. I mean this
25:00
25 minutes
same sort of thing. So I'm I'm really interested in how does AI mediate our norms, our behavior? Does it reinforce
25:07
25 minutes, 7 seconds
the gendered inequalities we see wider in society? And then when you kind of see that picture, it's pretty frightening. Um the big caveat is that
25:16
25 minutes, 16 seconds
this research snapshot focused on harms for a reason. I I I acknowledge that there are are huge opportunities um and
25:24
25 minutes, 24 seconds
that is something that we're hoping to go into for the for the fall on white paper, but for me it was really um present and timely and important that we
25:33
25 minutes, 33 seconds
started with the harms because ultimately um you know do no harm um kind of a harmvention approach is really really important particularly when we're
25:41
25 minutes, 41 seconds
in this era of backlash and regression in some instances um which mean I think we need to keep the the kind of focus on the harms and and maybe not gloss over
25:50
25 minutes, 50 seconds
them before we jump to the opportunities.
25:52
25 minutes, 52 seconds
I I couldn't agree with you more and I think what's exciting in this space is the the growth in data and research that's that's documenting the kind of
26:00
26 minutes
harms you were talking about. to the the representational harms or what it means for these gender stereotypical um large language models and natural language
26:09
26 minutes, 9 seconds
processing and um virtual assistant um chat bots that are very feminized in terms of the dating space which is
26:18
26 minutes, 18 seconds
something I'm day-to-day less familiar with but there's a fantastic new study released by the behavioral insights team and funded by Mindaroo which talks about
26:25
26 minutes, 25 seconds
the extent of tech facilitated gender based violence on dating apps so I'm I'm excited to see this massive massive growth in knowledge and data about the
26:34
26 minutes, 34 seconds
harms and and really agree with you then once we know enough about the harms then we can start to talk about the response.
26:40
26 minutes, 40 seconds
Yeah, absolutely. And actually so I do want to move to some questions to you too because you've been working on the gender legislative index for years now
26:49
26 minutes, 49 seconds
but it's recently ramped up and you've got this uh specific focus on artificial intelligence and technology for part of that work. I am really interested, you
26:58
26 minutes, 58 seconds
know, you've assessed some of these laws and how they are responding to gender.
27:01
27 minutes, 1 second
How how responsive to gender do you think think that Australia's current approach is to AI?
27:07
27 minutes, 7 seconds
Well, I think you've already hinted at this, Elise, which is the risk when we regulate in certain spaces or focus on certain issues, say non-consentual image
27:15
27 minutes, 15 seconds
based abuse, and don't have a holistic picture. So, my short answer would be not nearly gender responsive enough. um you know through the gender legislative
27:23
27 minutes, 23 seconds
index the team at UTS and I are asking this question whether laws adequately respond to the different ways in which
27:30
27 minutes, 30 seconds
legislation is lived by men women and gender diverse people and so we don't only ask what is this law trying to achieve but whose experiences are
27:38
27 minutes, 38 seconds
visible in the legislation what obligations does the law create what remedies will genuinely be accessible and what will the law's likely effect be
27:47
27 minutes, 47 seconds
and now we are applying this to AI and it's I think it couldn't be more timely.
27:51
27 minutes, 51 seconds
We're looking at over 80 laws related to online safety and AI. And it's not that Australia is devoid of regulation. We have fragments of legislation, but I
28:00
28 minutes
think it's largely gender blind. So, we don't have a a comprehensive economywide AI law. And so, instead, our legislation
28:08
28 minutes, 8 seconds
is spread across online safety, privacy, anti-discrimination, and criminal law.
28:12
28 minutes, 12 seconds
And it's it's quite similar to other jurisdictions that have these sporadic approaches rather than an omnibust sort of act. And and when I talk about it
28:20
28 minutes, 20 seconds
being gender blind, you know, it's really problematic when legislators, lawmakers treat a subject as gender neutral the same way some people treat
28:29
28 minutes, 29 seconds
tech as gender neutral in ways that ignore evidence that the harm, its severity, and the barriers to actually
28:36
28 minutes, 36 seconds
obtaining redress are all going to be shaped by gender. You know, we know women and girls are disproportionately targeted by synthetic sexual abuse, by
28:44
28 minutes, 44 seconds
misogynistic harassment, by other forms of tech facilitated violence, by stereotyping generated by AI, and by
28:51
28 minutes, 51 seconds
refusing to name women and girls as at particular victims of these harms at particularly vulnerable and at risk. It doesn't make the law fairer by treating
28:59
28 minutes, 59 seconds
everyone the same. It makes those most affected less visible. So here in Australia, I think it's often celebrated that in 2024, we have a federal criminal
29:07
29 minutes, 7 seconds
law recognizing non-consensual sexual material. And in that law, it doesn't matter whether the material is real, altered, or entirely fabricated.
29:17
29 minutes, 17 seconds
It's all captured, which is great. But criminalization is only one part of an effective response. And if you talk to someone affected by non-consentual
29:24
29 minutes, 24 seconds
image- based abuse, prosecution can take months or years. An image can be circulated in minutes. And you know, as
29:31
29 minutes, 31 seconds
a lawyer who loves and respects the law, sometimes the law is just too slow.
29:35
29 minutes, 35 seconds
And so in a in a legal response where you want to be asking, can you get the material removed faster, can you do something about when that
29:43
29 minutes, 43 seconds
material is re-uploaded, whether the victim's identity is protected when they complain so there's not retaliation, whether specialist support services for
29:51
29 minutes, 51 seconds
them, do they have access to some kind of legal aid as well as other sort of psychosocial support. So all of this sort of comprehensive approach I think
29:58
29 minutes, 58 seconds
is missing. And I think there's huge excitement around the proposed digital duty of care. And I definitely support the shift. It represents this shift to
30:07
30 minutes, 7 seconds
not waiting until the harm occurs and placing the burden on the individuals to pursue a remedy, but placing the burden on the internet service providers and
30:15
30 minutes, 15 seconds
the AI providers and platforms. But in my view right now, the draft bill which was tabled in September of 2026
30:24
30 minutes, 24 seconds
is gender blind. And and I can give you an example. The bill speaks to encouraging hostility towards women and
30:31
30 minutes, 31 seconds
gender equality and that sounds really exciting but the bill only makes that a problem if it's harmful to children.
30:38
30 minutes, 38 seconds
It doesn't worry about that language as being problematic for the whole of society for women and so it's in that sense it's really gender blind and an
30:45
30 minutes, 45 seconds
entirely missed opportunity to tech facilitate a gender based violence and a gender responsive reform. Um, you know, Australia wants leadership in the
30:53
30 minutes, 53 seconds
digital space because we were the first to lead with the social media ban and I think we could have leadership, but we really need to see this bill turned around to be something far more gender responsive.
31:02
31 minutes, 2 seconds
Yeah, absolutely. I mean, I think a lot of the discussions that I've been part of around this really talk about, you know, all of the permissive stuff, the
31:10
31 minutes, 10 seconds
lawful but awful, you know, that's not going to be captured by that that law.
31:14
31 minutes, 14 seconds
And so there's there's still so much um that's needed there. And then you know conversations with youth advocacy groups have been really concerned about well
31:22
31 minutes, 22 seconds
what does it mean you know with the prevalence of let's say deep fake pornography for instance amongst school children where the perpetrator and the
31:30
31 minutes, 30 seconds
victim are both minors basically they they are not um adults in the islands of the law and therefore you know is a
31:37
31 minutes, 37 seconds
criminalization approach really what we want in that first instance and I know we've got some members from the department of education who've joined this call and that's that's great to see
31:45
31 minutes, 45 seconds
you there but I think this is a big challenge that schools and the education system are grappling with like how do we actually infiltrate not just the the the
31:53
31 minutes, 53 seconds
legislative side but also the social norms side about this is you know acceptable use of AI and this is not acceptable use of AI and um anyway I
32:01
32 minutes, 1 second
think there's lots more conversation to be had there well I think I mean it's an incredibly important example in in a in a country that has far too young an age for
32:09
32 minutes, 9 seconds
criminal responsibility when you have a minimum age of criminal responsibility of 10 and you've got a child in late primary school or high school making a
32:18
32 minutes, 18 seconds
deep fake. This is something that will sit with them forever. And I don't think we can compare that harm to someone who's an adult who should have greater knowledge and be doing that at at scale.
32:28
32 minutes, 28 seconds
And so it is a really real conversation.
32:30
32 minutes, 30 seconds
And when you talk about the role of the law, people are quite um lawmakers, legislators, legal drafters can be quite narrow in what we expect of legislation.
32:39
32 minutes, 39 seconds
And yet there are countries such as Argentina and the Philippines who have online safety laws that actually talk about education that talk about gender
32:47
32 minutes, 47 seconds
equality and tech facilitated gender- based violence as being a core part of what everyone in society should be educated about. And so I think we we
32:55
32 minutes, 55 seconds
want to bring that education conversation into law making and expand, you know, what we expect um were
33:02
33 minutes, 2 seconds
protected against, but also what rights are protected in in our laws trying to address this. M yeah no absolutely um in
33:11
33 minutes, 11 seconds
terms of Australia I mean we've been seeing a lot around data centers recently um there's often a bit of focus on productivity gains at least for
33:19
33 minutes, 19 seconds
certain segments of the population what do you see as the least talked about but maybe most urgent um areas of AI and I
33:27
33 minutes, 27 seconds
guess do you have any policy recommendations or what's on your recommendation wish list I mean it's interesting if you ask what's least talked about it's it's
33:36
33 minutes, 36 seconds
something you need to dig through because I think right now this discussion is very much gravitating towards these two extremes around AI. So
33:44
33 minutes, 44 seconds
one there's this existential fear that AI is going to ruin the world. You know we've got these r these agents that are going to destroy everything that we take
33:53
33 minutes, 53 seconds
as as a given on the one end and on the other you've got this narrative around AI promising us faster and more efficient and more productive workdays.
34:01
34 minutes, 1 second
And I'm a user of AI and and a believer in its potential, a huge believer. But I think those conversations at the extreme distract us from something more in the
34:10
34 minutes, 10 seconds
middle which worries me more which is what is AI already doing to people whose lives are already shaped by unequal
34:18
34 minutes, 18 seconds
institutions. You know that's that's us that's that's this is where the gender question comes in. It's the everyday that that worries me. And I think as a
34:25
34 minutes, 25 seconds
result much of the law and policy debates are focused on bias.
34:29
34 minutes, 29 seconds
Is the data set representative? Well, no it's not. We know this for a long time.
34:33
34 minutes, 33 seconds
Carolina Creat's book told us if we needed evidence, does the algorithm discriminate? Well, very likely it will.
34:39
34 minutes, 39 seconds
And can we remove gender from inputs and and make the tech fairer? And yes, we probably can. And I think those questions are important to ask. But I don't think that's enough.
34:48
34 minutes, 48 seconds
And I think this focus on bias means we forget the structural nature of this problem. And I think recruitment is a really good example. So you can take an
34:56
34 minutes, 56 seconds
AI tool used as an assisted employment decision-m tool and remove remove names
35:04
35 minutes, 4 seconds
um Elise and Ramona being fairly um female identifying names, remove other gender markers, but the system will
35:11
35 minutes, 11 seconds
still probably reward a CV with uninterrupted full-time employment and it will still downgrade a CV where there's histories of part-time work and
35:19
35 minutes, 19 seconds
caring responsibilities because all of this is happening in an unequal society.
35:24
35 minutes, 24 seconds
And so for me, I would love to see the conversation pay a bit more attention to less about debiasing, debiasing the data and debiasing the algorithm and more
35:32
35 minutes, 32 seconds
about substantive equality and how can we take this as an opportunity to reflect on building an AI tech world
35:40
35 minutes, 40 seconds
that operates with this broader goal of substantive equality in mind because if not this these automated systems will produce inequalities at scale across
35:49
35 minutes, 49 seconds
employment and education and credit and healthcare and and and the list goes on.
35:53
35 minutes, 53 seconds
So to me that's that's a really silenced aspect and I think you know something that's a little bit skewed in terms of my policy wish list. You
36:01
36 minutes, 1 second
know, I I I guess I would go back to the digital duty of care because I think it's the opportunity now. And if I could say three things about the bill, I would
36:09
36 minutes, 9 seconds
love to see the bill once it's tabed in parliament actually name tech facilitated gender- based violence in the law as a problem and define it as a
36:17
36 minutes, 17 seconds
problem that comprehensively and and name Cedor obligations. Nearly every country in the world has has ratified CEDOR. So all of those 189 countries are
36:27
36 minutes, 27 seconds
obliged with what Cedor has said around the gendered harms of AI. The second thing I'd love to see is um you know platforms are often told do a risk
36:35
36 minutes, 35 seconds
assessment and and we've seen this in other spaces where a risk assessment allows big tech and corporations to tick
36:42
36 minutes, 42 seconds
a box and have a compliant risk assessment but they're not being asked to do gendered risk assessments. So if your research and and others is showing
36:50
36 minutes, 50 seconds
the gendered harms risk assessments must be gender responsive. We need to aggregate data across disability and
36:57
36 minutes, 57 seconds
race and indigenity and sexuality so we know the compounded effects on a diversity of women and girls for gender responsive risk assessments. And then
37:06
37 minutes, 6 seconds
third, can we deliver remedies at the speed of digital harm? And I think all you need to do is talk to someone who's experienced harm, especially online
37:14
37 minutes, 14 seconds
harm, especially in the form of a deep fake. Not that we want to narrow the issue, but violent or harassing comments. um partners using these tech
37:22
37 minutes, 22 seconds
platforms to perpetrate abuse or coercive control and non-consentual image based abuse the the the harm is
37:29
37 minutes, 29 seconds
immediate and it's escalating and so we need remedies that that operate at the speed of digital harm.
37:36
37 minutes, 36 seconds
Oh, 100% agree. And actually what you said earlier about kind of debiasing AI, it reminded me a little bit of, you know, um kind of, you know, when people
37:45
37 minutes, 45 seconds
would say, oh, you know, I don't see race, you know, it's that same sort of thing like, well, it's still there.
37:50
37 minutes, 50 seconds
[laughter] So, what are we actually going to do? And actually, we want to both acknowledge and look at what, you know, ways we can um combat unequal or
38:00
38 minutes
injust systems. So, you know, it's not about stripping that out and stripping the visual of that. So, yeah, I think
38:06
38 minutes, 6 seconds
that's um an apt comparison there. Um I'm conscious of the time and I would love to hear from our audience. But did
38:14
38 minutes, 14 seconds
you um why don't we see if there's any questions and then once we take some questions, we'll we'll maybe offer our
38:20
38 minutes, 20 seconds
um really active I can tell audience um some some tips. Um so that sounds did
38:28
38 minutes, 28 seconds
[clears throat] you want to start with the second one?
38:31
38 minutes, 31 seconds
um which is you know what are schools doing to address the harms of AI? Do you think there's enough going on there first and then I can I can add
38:39
38 minutes, 39 seconds
Yeah. Yeah. Absolutely. So on on schools so my research just recently didn't look specifically on what different schools
38:46
38 minutes, 46 seconds
approaches are. I do know from working with various youth advocacy organizations and some cross government that there is general um fear and
38:55
38 minutes, 55 seconds
there's been attempts at different ways of combating some of the harms that we're seeing particularly around tech facilitated gender based violence and
39:03
39 minutes, 3 seconds
deep fakes. I mean some of you may have seen uh state of man on the ABC by Mark Fenel starting to go into some of this and talking to what what's going on in
39:12
39 minutes, 12 seconds
schools. Um in general though from what I've seen from the research we are looking at a siloed and ad hoc approach
39:19
39 minutes, 19 seconds
still um what I haven't seen as much on is a basically a national conversation like what are what what kind of
39:27
39 minutes, 27 seconds
Australia are we hoping to create and how do we integrate these new technologies and whether it's this version of technology that we currently see or anything that comes in the future
39:36
39 minutes, 36 seconds
how are we establishing some of those norms around acceptable and unacceptable use noting um you In the civil and
39:44
39 minutes, 44 seconds
military um domains, we've got this thing called dual use technology.
39:49
39 minutes, 49 seconds
Basically, anything that could any technology that could be used for for good and for bad purposes is this kind
39:56
39 minutes, 56 seconds
of or not bad. That's you know that's a bit simplistic but could be for a civil uh use could be for military use. Now
40:04
40 minutes, 4 seconds
that generally has quite specific guidelines around that tech. I'm not necessarily seeing that for many AI
40:11
40 minutes, 11 seconds
applications this dual use right so it can be used for positive in so many ways and again going back to that um kind of
40:18
40 minutes, 18 seconds
our our understanding of technology is something that you know um impacts on the nature and the severity or pre
40:26
40 minutes, 26 seconds
prevalence of harm um I I think that there's still a lot of space in there that we are missing um and and I'd love
40:34
40 minutes, 34 seconds
to have further conversations for anyone who is in the education space and who wondering what to do. I think there's a there's an enormous opportunity there.
40:42
40 minutes, 42 seconds
Absolutely. And and UTS a couple of weeks ago hosted the Australian Public Policy Institute's annual summit, the policy forum um here and there were lots
40:50
40 minutes, 50 seconds
of people from the Department of Education involved. You know, I think first of all, I would add to to everything Elise has said that we live in a country where gender equality as
40:59
40 minutes, 59 seconds
something we teach in schools is not first principles. We're not teaching it from from the from kindergarten through to year 12. this idea of misogyny, what kind of society we want, women's rights.
41:10
41 minutes, 10 seconds
We don't live in a country where human rights language is part of the the the day-to-day norm. And so I think we are playing catch-up. You know, consent
41:18
41 minutes, 18 seconds
education comes far too late um and in many schools and it happens a day or two. It's not something embedded
41:25
41 minutes, 25 seconds
throughout. So suddenly we're now talking about risks for young people online. uh young people at a very early age being exposed to to pornography
41:33
41 minutes, 33 seconds
that's not always um sexually empowering but often harmful pornography at an age that's not age appropriate. And so I
41:41
41 minutes, 41 seconds
think we need to have, you know, really serious conversations about what guidelines are going to to govern our schools. In New South Wales, the the Minister for Education is proposing a
41:50
41 minutes, 50 seconds
reduction in the amount of time children spend online. And I think it's an entry point. So it it it's it's coming from a place of our students are not as well
41:58
41 minutes, 58 seconds
educated as their peers based on PISO scores. But if that's the entry point to have proper conversations about tech because it's not just about how what is
42:06
42 minutes, 6 seconds
the best learning opportunity but also what are we exposing children to when they're freely using these online tools.
42:11
42 minutes, 11 seconds
I think we use those entry points to to have those those conversations as well.
42:16
42 minutes, 16 seconds
Tanya um Kovac has joined us and she's asked an excellent question which is to to paraphrase Tanya if you don't mind um
42:24
42 minutes, 24 seconds
it's really easy to pay lip service to gender these days um and we want to have really serious conversations about gender you Tanya I'm happy to jump in
42:32
42 minutes, 32 seconds
first and say when I made the submission to the digital duty of care a dear colleague of mine said I definitely know all about your work I promise but I
42:41
42 minutes, 41 seconds
finally understood what you mean by gender responsive legislation when I read your submission and it got down to the specifics of we need to add this
42:49
42 minutes, 49 seconds
word into the draft bill. We need to define these terms in this way. We need to use theor as a governing framework the same way we use the children's
42:57
42 minutes, 57 seconds
rights convention in several pieces of Australian legislation. So I think there's one option to get beyond the the lip service agenda is to make the
43:04
43 minutes, 4 seconds
conversations as substantive and evidence-based as possible. And that's why the team here as part of the gender in the law program is also trying to dig
43:12
43 minutes, 12 seconds
up examples of how other countries are responding with really specific provisions from laws so that it's not easy to for the for policy makers or
43:21
43 minutes, 21 seconds
legislators to say oh we're doing this um when we know they can do it better in a really um a much more substantiated way. Elise, did you have anything you
43:28
43 minutes, 28 seconds
wanted to add to that? Yeah, I I just I mean I I completely understand and um empathize. You know, I feel a lot of
43:36
43 minutes, 36 seconds
solidarity and um you know disappointment for the current system.
43:40
43 minutes, 40 seconds
You know, we have had major defunding of some of the innovation programs that could have been an open door to help get you know more women and gender diverse
43:48
43 minutes, 48 seconds
and first nations folk into this sector and start to make an impact. you know, just more generally, not speaking to AI
43:55
43 minutes, 55 seconds
startups specifically, but in Australia, um, uh, all funding to female only um,
44:03
44 minutes, 3 seconds
teams, it was at about 2% of all venture capital funding that went to them, and that was down from a record uh, high of
44:11
44 minutes, 11 seconds
4% the year previously. So, we're at this really um, systematic stage where,
44:18
44 minutes, 18 seconds
you know, that's just in entrepreneurship. But then yeah, if you look at women's organizations and all of the rest, we we are seeing a bit of a hollowing out and and some of this is
44:26
44 minutes, 26 seconds
followed from the US and and what's happening with Trump. Um certainly in international development, we are seeing this. I think you know it's hard to know
44:36
44 minutes, 36 seconds
what the solution is in some ways um without those in power and with funding kind of taking this seriously and I
44:44
44 minutes, 44 seconds
think that that is the absolute call to action that needs to be made. I know we have only a small um pool of funding here in terms of philanthropic funding
44:52
44 minutes, 52 seconds
in Australia but wherever possible it should be used in really gender transformative ways and actually that is such a strong leverage point. Um
45:00
45 minutes
government does also have mandates around this. So the Australian federal government has guidelines and has had them since about 2022 2023 where all
45:09
45 minutes, 9 seconds
government departments have to um for any new policy proposal or cabinet submission it has to go through a gender um analysis. You know if it meets a
45:17
45 minutes, 17 seconds
certain threshold which uh Leonora professor Leonora Riss and I have argued AI anything to do with AI does it needs a more detailed gender impact
45:25
45 minutes, 25 seconds
assessment. Now, this should trigger mechanisms that provide adequate funding and resourcing and um kind of policy
45:33
45 minutes, 33 seconds
provisions that are gender responsive on this issue. You know, from my research so far, I've heard all kinds of stories that really worry me when it comes to
45:41
45 minutes, 41 seconds
that, including um you know, government either not doing it, saying that things like AI or other topics, well, they
45:48
45 minutes, 48 seconds
impact everyone and therefore they don't require agenda analysis cuz everyone's impacted. And you know this is just it's a fundamental misunderstanding. But then
45:57
45 minutes, 57 seconds
I've heard on the other hand where where government departments are uh you know required to do some of these um analyses
46:04
46 minutes, 4 seconds
you know they chuck it into AI and so you get whatever biases come out of that system. It might be good uh it might not be and also it doesn't replace the need
46:12
46 minutes, 12 seconds
for actually talking to women and talking to um experts in this sector too. And so the way I see it is that
46:19
46 minutes, 19 seconds
really we do need to partner up and I think you know Leonora and I you know certainly have actions um we have to keep this on the agenda. I think that
46:27
46 minutes, 27 seconds
there is you know more generally in society right now we're seeing a lot of polarization. I live in a small uh rural community in New South Wales. I
46:34
46 minutes, 34 seconds
certainly see it in my own community let alone um more broadly. Um so being able to make sure that gender is still on the table. I think what we see from global
46:43
46 minutes, 43 seconds
research is that, you know, when you get kind of closer to equality, you often um that's when gender equality is at most risk, right, of either stagnation or
46:52
46 minutes, 52 seconds
backsliding under this idea that it's already achieved or it's not an issue.
46:57
46 minutes, 57 seconds
Um some great research actually points to those who believe that gender equality has already been solved are most at risk of perpetuating inequality.
47:05
47 minutes, 5 seconds
And so for me, I think it's a real worry. I do see the defunding of women's orgs, innovation, AI for um those coming through who maybe look a bit different
47:14
47 minutes, 14 seconds
from um kind of the historic norm as as a priority for sure.
47:18
47 minutes, 18 seconds
Excellent, Elise. That was that was a great and compelling response. Um I'm just going to throw to one more question before I have a question back to you,
47:25
47 minutes, 25 seconds
Elise, which is Eliza's. Um, so Eliza's asked an excellent question which is, you know, so often AI generated content
47:32
47 minutes, 32 seconds
isn't going to fit into to a category of explicitness.
47:36
47 minutes, 36 seconds
And so then it escapes the law. There's an image created that another person doesn't want to be pictured in. And how do you police this? And and I think it's a really great question and a lot comes
47:44
47 minutes, 44 seconds
down to content moderation and gender responsive and culturally responsive content moderation. So just because something won't be explicit or labeled
47:52
47 minutes, 52 seconds
explicit and therefore falls into a harmful category doesn't mean it's not causing someone harm or doesn't mean it wasn't against someone's consent. And actually Australian legislation has a
48:01
48 minutes, 1 second
really interesting inclusion which is recognizing that when someone who would normally wear cultural dress is photographed without it that can be
48:09
48 minutes, 9 seconds
harmful content for them. Um because it does have huge repercussions including um repercussions culturally as severely as honor killings. And the fact that
48:17
48 minutes, 17 seconds
Australian law recognizes that is incredible. What we want to be seeing is platforms thinking about this when they create the tools in the first place, which is really what is often meant by
48:26
48 minutes, 26 seconds
safety by design or ideally equality by design where you start to think what might be the experiences of an individual of this platform and
48:34
48 minutes, 34 seconds
therefore should there be automatic tagging or is this something that someone should be able to consent to before an image that they haven't consented to appears online and in the public domain.
48:44
48 minutes, 44 seconds
Um, so great question Eliza. Um, if I can ask a question back to you, Elise, [laughter] which is really one for our audience, which is, you know, I think
48:51
48 minutes, 51 seconds
the Victorian Women's Trust has done a great job at bringing us together today.
48:56
48 minutes, 56 seconds
Um, really wants practical outcomes from their work. And so, if you could give some advocacy actions that you would recommend anyone listening could take
49:04
49 minutes, 4 seconds
and we have a huge diversity in the audience of experts and people clearly quite close to government and and people who may be experiencing AI in their
49:12
49 minutes, 12 seconds
unrelated, you know, sort of day-to-day lives. What actions could people take um to do something about these gendered harms?
49:20
49 minutes, 20 seconds
Yeah, I've got a few suggestions. So, first up, um there are opportunities like parliamentary submissions. Now, it sounds a little bit boring and a little
49:28
49 minutes, 28 seconds
bit nerdy, but hopefully you're all here because you care about policy or impact in some way. It can be a really effective way to ensuring that your
49:35
49 minutes, 35 seconds
views, your research, your organization's imperatives are reflected in policy. Um, let's not forget that, uh, you know, industry has very well-
49:44
49 minutes, 44 seconds
paid lobbyists who are able to make some submissions or do other lobbying efforts all the time. I'm really conscious that
49:51
49 minutes, 51 seconds
we need to band together as a as a, you know, community, um, and and put, uh, make input. So uh Ramona and I um
50:00
50 minutes
professor Leonora Ris the Victorian uh women's trust you know a group of other organizations the newani gender justice
50:07
50 minutes, 7 seconds
institute we all put in one recently on uh gender and artificial intelligence I welcome anyone who wants to write with me write to me and say hey can I join
50:16
50 minutes, 16 seconds
one of your submissions or I just saw this you know I I'm happy to participate but I think I need some more support whatever that that's one really great
50:24
50 minutes, 24 seconds
avenue um secondly I mean you the snapshot is out. What are your responses? What do you want to see done?
50:31
50 minutes, 31 seconds
Um, again, email, um, text, you know, find a way to communicate with us and share your thoughts and perspectives.
50:38
50 minutes, 38 seconds
Um, we will also be doing a roundt discussion workshop in Victoria, um, in November. Um, again, uh, write to us,
50:46
50 minutes, 46 seconds
um, see if we can squeeze you in, um, as a way to kind of provide some input to our white paper, which will go into more detail on kind of policy options that
50:55
50 minutes, 55 seconds
government and communities might be able to undertake. And I think more broadly about this, have conversations with your your family, your friends, your
51:03
51 minutes, 3 seconds
neighbors, your community, even those who maybe don't share the same perspectives as as you. I think some of this is a consciousness raising. It is a
51:10
51 minutes, 10 seconds
literacy campaign. It is a having a dialogue about what may [clears throat] or may not be known in the community. Um
51:17
51 minutes, 17 seconds
because ultimately I think I want platforms to be far more responsible as Ramona has said and I also want policy
51:24
51 minutes, 24 seconds
and legislative responses to be better but they will only ever um be as good as also our community. We have to change
51:33
51 minutes, 33 seconds
some norms around AI use and we have to have some conversations uh critical conversations not just with young people but all generations. Um, so that would
51:40
51 minutes, 40 seconds
kind of be my actions. Ramona, do you have any actions you'd um add for our audience?
51:45
51 minutes, 45 seconds
I I do. Although I think all of yours are great and really speak to what it means to live in a democracy. We we have to believe we can change the current
51:52
51 minutes, 52 seconds
state of play. Um, but I would give a second shout out to Nadia who in the comments mentioned the AI pledge, the inclusive AI pledge. So some of you may
52:01
52 minutes, 1 second
know Champions of Change did a panel pledge. So we weren't seeing male only panels anymore. And when you see a male only panel at a conference or an event,
52:09
52 minutes, 9 seconds
you can call it out. Well, the inclusive AI pledge is a pledge that you want the way in which we use AI every day will become more inclusive. And on the link
52:18
52 minutes, 18 seconds
that Nadia has shared, there's actually a prompt you can use. So, I use AI almost every day, I must say. And it allows you to tell your AI, well, I don't want you to bring out bias data.
52:27
52 minutes, 27 seconds
And when you write an agenda or suggest speakers or do a summary of an issue or or generate research for me, I want you to acknowledge that there are biases and
52:36
52 minutes, 36 seconds
to challenge those biases. And and I think it's also a self-empowering tool that gives us back some power in the hope that we can actually change the way
52:43
52 minutes, 43 seconds
the machine is learning. Um so a big shout out to the AI inclusive AI pledge.
52:49
52 minutes, 49 seconds
Um I think you want to ask harder questions when AI is introduced. So some of you might be working in workplaces that have introduced AI. um you might be
52:57
52 minutes, 57 seconds
forced to use an AI um if you have a new recruitment and they're using an employment decision-making tool or they're you're encouraged to use AI in
53:05
53 minutes, 5 seconds
the workplace to do research or to be more efficient or more effective. Ask the hard questions. Has this tool been tested for its gendered impacts? Will a
53:13
53 minutes, 13 seconds
person who's affected by a decision if you work in a bank or if you work in an education sector selecting um applicants? Will a person affected by a
53:21
53 minutes, 21 seconds
decision be able to challenge the decision that's been made by an AI? and who in your organization is accountable.
53:26
53 minutes, 26 seconds
I think it's important to try to find ways to ask these questions. And third, in addition to writing submissions, which I strongly endorse, although I
53:34
53 minutes, 34 seconds
fully appreciate how much work they involve, use your vote. You know, we have an election coming up in New South Wales in March the 13th and in Victoria
53:41
53 minutes, 41 seconds
next month. Ask candidates, what do you actually think about tech facilitator gender- based violence, about algorithmic dis discrimination, about big tech and corporate accountability?
53:51
53 minutes, 51 seconds
you know, don't just ask, which I think is an easy question, you know, what are you doing to support AI for innovation?
53:57
53 minutes, 57 seconds
What are they doing to support inclusive, accountable, and gender responsive innovation? So, I think being in an election period in a couple of
54:05
54 minutes, 5 seconds
states is an is a really um wonderful opportunity.
54:10
54 minutes, 10 seconds
So, at least you've already shouted out to a few things coming up. We'd love to hear more about the round table and the outputs. Um, and here at UTS, we have
54:17
54 minutes, 17 seconds
the gender and the law program, including a newsletter. So, do sign up to hear more about the upcoming opportunities in the AI space. Um, and I
54:25
54 minutes, 25 seconds
hope you're all going to be watching the digital duty of care bill as it progresses. A huge thank you again to the Victorian Women's Trust for
54:33
54 minutes, 33 seconds
organizing what I've thoroughly enjoyed um this excellent conversation with um Dr. Elise Stevenson. Um do grab a copy
54:41
54 minutes, 41 seconds
of Gender and AI, Where's the Harm? um and looking forward to the next opportunity to speak with you Elise on
54:48
54 minutes, 48 seconds
these really fundamental issues. Thanks very much. Much for hosting. Thanks all.
54:54
54 minutes, 54 seconds
[music]
54:58
54 minutes, 58 seconds
[music]
55:05
55 minutes, 5 seconds
[music]

Share