Breaking biases: Unveiling persistent echo of 'shrink it and pink it' in AI, championed by women for change

In a realm where innovation reigns supreme, the ghost of an old adage still whispers in the wind: "Shrink It and Pink It." Once a dictum for product design, it now silently weaves its threads into the very heart of Artificial Intelligence (AI), blurring the lines between antiquity and advancement. As we stand on the cusp of an AI-driven era, the question looms large: How does a belief rooted in yesteryears remain steadfast in the face of technology's rapid evolution?

Behold the convergence of past prejudice and present progress, where the hues of pink and the art of "shrinking" transmute into digitized dogma. The same sentiments that once pigeonholed women's interests into dainty boxes now insidiously taint AI responses, entangling our digital future with historical biases. It's a cautionary tale of how the tides of change don't always wash away the footprints of the past.

In this exposé, we embark on an exploration that transcends mere problem identification. We voyage from the origins of "Shrink It and Pink It" to the disturbing underbelly of AI responses that mirror the sexism of the past. But here, our journey isn't one of despair – it's a beacon of hope. We're peeling back the layers to showcase the stories of the women who've stormed the bastions of technology and AI, carving their names on a landscape once dominated by a singular gender.

A bleak outlook for women

Gender bias in AI is an ominous cloud overshadowing the realm of artificial intelligence. It threatens to deepen the longstanding gender divide. Alarming reports worldwide paint a grim picture of how AI could adversely affect women.

• A recent McKinsey report rings the alarm: The AI surge might disproportionately displace women from the workforce by the decade's end. American women, in particular, could bear the brunt of AI-induced job losses, exposing a stark gender bias in AI's impact distribution.

• The World Economic Forum (WEF), a global gender equality advocate, delivers equally sobering news. The Global Gender Gap Report 2022 predicts gender equality to remain a distant dream, with a staggering 132-year wait. Progress, albeit present, crawls at a frustratingly slow pace.

• Beneath these troubling statistics lies a glaring underrepresentation of women in AI and tech. UNESCO's 2019 data reveals that merely 12% of AI researchers are women. In the software development realm, women occupy a mere 6% of the workforce. Shockingly, women are 13 times less likely to file ICT patents compared to their male counterparts.

These reports unveil a bleak narrative—a future where AI could amplify gender disparities. The question that resounds through these findings isn't just about AI's capabilities but the ethical responsibility to ensure AI's advancement doesn't further marginalize women.

When code reflects culture's shadows

Picture this: a world where technology, that most modern of marvels, not only listens but also perpetuates the same age-old stereotypes that have plagued society for generations. It seems like a paradox, right? But as AI models become more sophisticated, it's becoming increasingly evident that they are not immune to the biases ingrained in our culture. Let's dive into the world of AI and gender bias, where virtual responses echo real-world inequalities with uncanny accuracy.

In the face of such revelations, it becomes clear that AI's journey towards objectivity is far from over. These instances of bias, whether in text, imagery, or responses, are a stark reminder that technology mirrors society's shadows.

Women leading the charge against gender bias in AI

In the rapidly evolving landscape of Artificial Intelligence (AI), gender bias has emerged as a critical concern. Three remarkable women in the AI domain, Dr Fairoza Amira Binti Hamzah, an AI Engineer and AI Ambassador in Malaysia; Shruti Nagpal, a Research Scientist at Sony AI renowned for her contributions to AI ethics and creativity; and Dr Begüm Genç-McEvoy, a Data Scientist and Ambassador for Women in AI Ireland, have provided valuable insights.

Dr Begüm Genç-McEvoy emphasises that AI models inherit biases from human-generated data, stating,"These models inherit the sexist and anti-feminist responses that are introduced by humans, who are the source of the data."This includes biases such as sexism and anti-feminism.

Dr Fairoza Amira Binti Hamzah highlights the importance of training data, which can introduce gender bias into AI responses. She points out that using female names and voices for AI assistants like Amazon's Alexa and Apple's Siri can further reinforce stereotypes about women, stating,"The data and labelling often exhibit bias, leading the model to learn from incorrect data, ultimately resulting in erroneous predictions."

Addressing biased AI responses, Dr Begüm Genç-McEvoy calls for a balance between innovation and ethics in AI development. Shruti Nagpal adds that AI models are continually evolving, emphasizing that"AI models are a work in progress."

Dr Fairoza Amira Binti Hamzah stresses that combatting anti-feminist and sexist AI statements requires more than user feedback. She suggests retraining AI models on diverse datasets and educating users about AI bias.

Dr Shruti underlines that"AI is a potent tool, provided it is used appropriately and responsibly."She emphasizes that AI-induced gender bias can significantly hinder global gender equality, affecting various aspects of life, from job opportunities to loan approvals.

Regarding advancing gender parity, Dr Fairoza Amira Binti Hamzah notes that there are more female AI researchers than developers in Malaysia due to academic preferences. Dr Genç-McEvoy highlights the need to address the decline of women in higher-level positions and recommends creating female-only academic positions and targeted recruitment in companies.

In building inclusive AI models, Dr Genç-McEvoy encourages developers to adopt trustworthy AI principles and assemble diverse teams that question and research freely.

Dr Hamzah insists on"holding AI developers accountable for the ethical implications of their work, having a diverse team of developers working on AI systems, and incorporating fairness metrics into the design of AI systems"to ensure responsible AI evolution.

Dr. Hamzah also emphasizes the roles of governments, the private sector, and AI organizations in promoting gender equality in AI, highlighting the importance of policies, STEM education for girls, mentorship programs, and equitable hiring practices.

These visionary women illustrate that addressing gender bias in AI is a multifaceted challenge, but through education, diversity, and collaborative efforts, meaningful progress is attainable. They serve as inspiring catalysts, leading the way towards a more inclusive AI-driven society.

Crafting an equitable future

In the echoing chambers of AI, gender bias reverberates, threatening to cast shadows on the future. But in these echoes, there's a call to action—a promise of progress. To dismantle gender bias in AI, we must first recognize its existence, understand its roots, and acknowledge the role of data and developers in perpetuating bias.

The path forward isn't solely about combating bias but also about creating a technology landscape where everyone has a seat at the table. In this ever-evolving narrative, the conclusion remains open-ended. The question now echoes: Will AI become a realm of equality, or will it merely amplify the biases of the past? The answer lies in our hands, our code, and our commitment to progress. It's a future where the echo of "Shrink It and Pink It" is replaced by the harmony of diversity, equality, and innovation.

WATCH WION LIVE HERE