We are celebrating International Girls in ICT Day today, April 27th, 2023. I’ve seen numerous posts on social and traditional media throughout March and April showcasing women working in tech, including an impressive list of 100 women in the field of AI here in Norway. There has also been an array of events centered around women and technology, particularly focusing on the AI “revolution”. OpenAI’s CTO, Mira Murati, has appeared in the media as an outspoken proponent of regulating AI.
While I celebrate most of these initiatives, I cannot shake a feeling there is at least some amount of “gender-washing” behind them. Currently, the field of AI, like much of the tech industry, is predominantly male-dominated with around 25% or less women in the workforce. This is problematic because it often overlooks the perspectives and experiences of women in the development and deployment of AI systems. As seen with other technologies, this can lead to biases and blind spots with serious consequences.
Artificial intelligence (AI) is advancing rapidly, with undeniable impacts on our society. From self-driving cars to personalized assistants and “compassionate AI friends”, AI is revolutionizing how we live, work, and interact. As we develop increasingly advanced AI systems, it is crucial to consider the people working on them and the values they bring.
Recruitment systems using AI have shown bias against women, while facial recognition technology has demonstrated higher error rates for women and people of color. This is likely due to insufficiently diverse datasets used to train these systems. Such biases have led to misidentification and wrongful arrests, causing significant harm to individuals and communities. To address these issues, we need more diversity in AI at all levels, from research to testing, deployment, and regulation.
Women provide different perspectives and experiences that can help develop and deploy fair, ethical, and inclusive AI systems. However, it’s not just about gender equality; it’s also about ensuring our survival as a species by incorporating women’s perspectives, experiences, and goals in AI systems. Developing AI with traditionally feminine traits, such as empathy, compassion, and care, is vital for aligning AI systems with human values and priorities that promote our thriving as a species. We need AI that goes beyond merely streamlining work tasks and hyper-productivity.
The irony and gravity of the situation were evident in a March 5th “60 Minutes” episode about AI chatbots like ChatGPT, featuring interviews with Microsoft President Brad Smith and AI ethics research scientist Timnit Gebru. Timnit was fired from Google in 2021 after co-writing a now-famous paper on the dangers of large language models and later co-founded The Distributed AI Research Institute, an organization that calls itself “a space for independent, community-rooted AI research, free from Big Tech’s pervasive influence”. Brad Smith and his colleagues at Microsoft tried to brush off questions on why their chatbot’s “alter-ego” Sydney had a nervous breakdown and tried to get a New York Times reporter to leave his wife. Timnit does a great job explaining the flaws of large language models and giant datasets, watch the “60 minutes overtime” segment for the full interview.
The episode highlights not only the dangers and unknowns surrounding AI but also the need for diverse perspectives and voices in AI development and deployment, as well as more representation in the datasets these large language models are trained on.
While Microsoft focused on integrating OpenAI’s technology with its products using the “new and improved” GPT-4 model, the DAIR Institute held its first “Stochastic Parrots Day” on May 17th, reflecting on the warnings they had issued. Some harms, such as exploited workers filtering hateful content and the mounting environmental costs of LLMs, have already occurred, and the risks may only be accelerating. Not to mention the Google engineer claiming chatbots are sentient,
The tech industry’s yin-yang, masculine-feminine dynamic is evident in the way US startups create disruptions and get eventually bought by big tech, which then spreads the disruption further. The EU often steps in to regulate and set the rules of the game, as seen with the growing list of fines, GDPR, Schrems II, and the upcoming EU AI Act.
In conclusion, we cannot afford to overlook women’s perspectives and experiences in technology. This year’s Girls in ICT Day theme, “Digital Skills for Life”, aims to inspire and empower girls to pursue a future in ICTs. Achieving their goals also means including the diverse perspectives and experiences of women and other underrepresented communities in the development, deployment, and regulation of AI systems and technology. By involving people from all walks of life, we can ensure that AI systems are developed and deployed ethically, fairly, and inclusively. This is crucial for shaping the future of our society and our survival as a species.
Mix-master – the text content is prepared by human hands, using minimal assistance from AI tools (spelling, grammar).
Leave a Reply