Gender Politics of Artificial Intelligence

By Bonnie Chiu

Meet Tay. She is a Microsoft Artificial Intelligence (AI) chatbot launched on March 23, 2016, in the month of Women’s History Month. In less than 16 hours, Tay was taken down as it “tweeted wildly inappropriate and reprehensible words and images”, according to Microsoft. A few media outlets investigated Tay’s tweets and Quartz noted that the bot had itself started spouting racist, sexist, anti-Semitic comments.

I learnt about Tay’s case as a participant of the Gender and Technology Institute in Malaysia in April this year, and I was appalled. As an occasional consumer of sci-fi movies, with AI being a hot topic in these movies, I never before came across such critical perspectives seen through gender lens.

While technology is neutral, the application of technology can never be.

Our offline biases are reproduced on the online space. In fact, some of the extreme views that are politically incorrect to express in person, are condoned on the Internet. Most of us have heard of ‘trolls’ (people who discord on the Internet by starting quarrels or upsetting people, by posting inflammatory, extraneous, or off-topic messages in an online community, as defined by Wikipedia). In my circle of activists, all have had experience being targets of trolls. The example of Tay the chatbot shows that AI can be easily manipulated by trolls, leading to online hate speech that disproportionately affects certain groups in society.

Learning about AI also made me think about how AI links to our visual archive at Lensational — images produced by women whose voices are currently not heard on a global stage.

Simply from a volume perspective, men outnumber women in producing content shared on the Internet. Currently, 200 million fewer women than men own mobile phones.

According to a recent WIRED article, a study that used two well-known image banks highlighted a ‘predictable’ gender bias in their depiction of activities, ‘teaching’ the machines a sexist viewpoint. Images of domestic activities, such as shopping and washing, featured predominantly women, while those showing coaching or shooting featured men. In this case, computers then began to associate people in kitchens with women, even if the subject was male.

Perpetuating sexism and other harmful beliefs in our society is a danger, and it sheds light on who is currently producing these algorithms. Do we need to think more carefully about who controls and owns these means of production, and who they are accountable to?

Not to mention the possibility of using AI-powered weapons and other dangerous implications that have been summarised in an open letter by Stephen Hawking, Elon Musk and other experts.

However, these discussions must not be contained within the techies and geeks, but must involve the wider society. Today in London saw the opening of the Glass Room, a pop-up exhibition designed as a tech store that challenges the public to reconsider the technologies we use every day and what choices we can make in our quantified society. Most AI and technology debates look at their economic implications, but it is equally important to consider their social and political implications.

This blog post is written as part of the Millennial Blogger series, this month looking at the ethics of AI.