Artificial intelligence is a hot topic. It's one that breeds all manner of questions: can AI ever be sentient? Why is everyone talking about chatbots right now? Will AI take my job? But there's one specific question has been on my mind lately, given that the theme for International Women's Day 2023 is ‘DigitALL: Innovation and technology for gender equality’: will AI help, or hinder, the pursuit of gender equality?
Artificial Intelligence has shown troubling signs of bias, the work of Safiya Umoja Noble’s Algorithms of Oppression (2018), showed how seemingly ‘impartial information sorting tools’ actually perpetuate systematic racism. A new study by LCFI researchers has found that films entrench gender inequality in AI. Researchers from the University of Cambridge argue that such cultural tropes and a lack of female representation affects career aspirations and sector recruitment, and that without enough women building AI there is a high risk of gender bias seeping into the algorithms set to define the future. Last month, The Guardian found that AI tools rate photos of women as more sexually suggestive than those of men, especially if nipples, pregnant bellies or exercise is involved. So far, so scary.
An issue with AI and misogyny at the forefront of my mind is synthetic porn, a new and booming industry. Anyone with access to a computer can now create and sexualise their ‘perfect’ woman - something that is adding to the ever growing dehumanisation of women. AI porn feels dystopian, and it's a form of misogyny that most of us can't quite grasp. Technology is a product of humans and society, and this society is one full of prejudice - so how can we expect AI to help us achieve gender equity, and not worsen it? Is that even possible?
Despite all these concerns, some experts believe that AI really can help feminism, not hinder it.
Speaking with GLAMOUR, Lacey Hunter, the CEO and Co-Founder of TechAid, says that the question of pornograpy and human trafficking as it pertains to AI is multifaceted and far from settled: “On its surface, this might seem like just another iteration of objectification and exploitation of the female form. However, increased precision and speed of video meta-analysis made possible with AI tooling may enable more efficient cross-examination of web content to both identify and report bad actors, as well as to identify and assist victims or folks seeking support.”
Lacey believes that artificial intelligence can be used in numerous ways to support women and achieve equality, and that progress is already happening along multiple dimensions, citing AI innovation in the health and medical data space. “Shyro, a female-founded application that uses AI tooling to bring health searches (such as for fertility clinics, pre/post natal care and others) into a private instance that cannot be accessed by third parties, and matches users with potential tools, deals and discounts. Shyro disinter-mediates intrusive ads from the likes of Facebook and Google that use meta-targeting to ask for inherently sensitive, personal information and further run the risk of exposing this information elsewhere - as womens’ reproductive rights remain far from certain around the globe, security and privacy of related information remains paramount," explains Lacey.
Lacey tells me about LifeForce, another AI innovation specifically designed to support women, mothers and single mothers. It is an AI-enabled platform designed to match the supply and demand of aid subsequent to Russia’s invasion of Ukraine: “The initial product design took the needs of women and especially single mothers into account after the conscription of males into the Ukrainian resistance effort, as did the design of its analytics, the prioritization of product creation, and the ticketing system for assistance with delivery and community driven help. Another low-tech, but hugely impactful detail was also kept in mind - shelters were designed and built with children's play areas included.”
Brianne Kimmel - the Founder of Worklife Ventures and the most active female investor in Generative AI in the world today - believes AI is the greatest force for equality and democratic ideals we've ever known. Brianne does not believe that gender biases in AI are unavoidable, telling me that she's seen many tools that help people, especially men, be more empathetic and give women an equal voice in meetings: “Poised, an AI-powered meeting assistant, gives you real-time metrics like interruptions, filler words, and time spent talking during a meeting. This is especially helpful for managers to empower and call on early career professionals to share their ideas in meetings. We’re seeing a whole new wave of tools that help us be kinder in emails and other forms of written communication and give credit where credit is due. Promotions in the future will be based on contribution and merit, not just given to the loudest and most confident voice in the room.”
Dr Emilia Molimpakis, Neuroscientist and the CEO & Co-Founder of thymia explains how the very problem of bias in AI, may help us create real-world solutions.
Dr Emilia explains that while AI can be an incredibly powerful tool, unless developed responsibly, ethically and with great care it can open the door to significant biases of all sorts, including gender biases. Saying “This is because AI cannot think for itself; it can only draw conclusions based on the datasets on which it has been trained. So if the data is gathered in a way that means it inherently contains biases, then this will lead to bias in the model as well. For example, in a model trying to estimate the prevalence of depression, if there are more women than men in the dataset or there is not an even split between women and men with high and low depression scores, or the researchers only include women with high scores and men with low scores, you can easily imagine how the model then assumes that any new woman it encounters will naturally be more likely to have a high depression score.”
Continuing to tell me that “Women are so often underrepresented within research and data - particularly in areas such as healthcare - leading many public data sets to be inherently biased against them in various ways. It is up to the researchers building the AI tools to continuously check for such biases in their datasets and in the models themselves in order to catch and correct for it. This is considered ethical AI and is best practice. The Alan Turing institute has published really excellent guidelines around ethical AI principles and if researchers follow these, they can easily overturn any potential gender bias in their models.”
"For the same reason that AI models can naturally create biases, they can also be used to help overturn any unfair preconceptions that have been socially ingrained in us as humans. If an AI model is fed a well-balanced dataset, based on a diverse group of men and women, all receiving equal pay and progressing equally through the ranks, this AI model will then learn and assume that any new woman or man it encounters should be treated equally to one another. AI could be used to breathe fresh air into what is a stagnant societal and workforce environment with respect to gender inequality and could help to show with more objectivity that women and men can and should be treated equally and offered the same opportunities."
As we live in the throws of an AI revolution, the most important thing we can do is stay informed. AI and its future is not something that has to happen to women. Maybe it could be for women?
