In this particular instance from March, the term “artificial knowledge” is utilized. (REUTERS)
Recent forecasts indicate that achieving gender equality may require 286 years if progress continues at the current pace. However, the uncontrolled proliferation of AI could lead to adverse consequences by perpetuating existing gender biases. The data fueling AI often mirrors prevalent stereotypes and gender biases, thus inadvertently perpetuating these biases. It is imperative for programmers to create tools that aid AI in identifying and mitigating these stereotypes.
This underscores the urgent necessity of involving more women in all aspects of AI development, encompassing STEM roles, education, research, and publications. Regrettably, the data paints a stark picture. Globally, women are significantly underrepresented in science, research, and advancement positions. Moreover, there exists a notable disparity in representation between male and female authors and experts in the AI domain, with only 18% of female attendees at esteemed AI conferences. The imbalance is further highlighted by the fact that merely 14% of authors of AI scholarly articles worldwide in 2020 were women.
A report by UNESCO highlights the global shortage of individuals working in AI-related roles. When AI systems are not developed by diverse teams, they are less likely to cater to diverse user needs and may not align with human rights principles. The underrepresentation of women in machine learning, programming, and technology supervision roles exacerbates this issue. With only 18% of undergraduate computer science degree holders being women and less than 25% of women in the tech industry holding senior management positions, the field remains predominantly male-dominated.
To address this challenge, there must be a concerted effort to increase female representation in AI, information technology, and software engineering teams. Educating male engineers about gender bias is essential to enable them to assess data, design choices, and computational decision-making through a gender-sensitive lens.
Gender biases in AI and machine learning systems can stem from feedback loops resulting from the lack of diversity in data technology. The example of ChatGPT’s recognition of biased language patterns underscores the risks faced by knowledge workers and factory workers alike due to AI. In this scenario, individuals and marginalized groups may feel further marginalized. For instance, a study by experts from Brazil’s Universidade Federal de Minas Gerais across 59 countries revealed biased image search results based on terms like “beautiful” and “ugly,” perpetuating racial and gender stereotypes.
Similarly, Amazon’s resume screening algorithms exhibited gender bias by downgrading resumes mentioning the term “women” or references to children’s schools. The flawed algorithm had to be scrapped as it relied on historical data that favored male applicants.
Despite the disheartening statistics, there are promising developments on the horizon. According to Coursera’s 2021 World Skills Report, women are enrolling in digital STEM courses at a higher rate post-Covid-19 crisis. The proportion of female Coursera enrollees rose from 38% in 2018–19 to 45% in 2020.
Enrollment of women in STEM courses teaching in-demand digital skills increased from 31% in 2018 to 38% in 2020. These increased participation rates persisted in 2021, with women accounting for 45% of overall course enrollment and 37% of STEM enrollment.
The pivotal elements of consciousness and intersectionality hold the key to addressing these challenges. While a global initiative is underway to establish an AI ethics center, it is imperative that this governing body deliberately considers and evaluates the gender implications of AI outcomes.
There is still an opportunity to counteract these trends, even though new AI systems have the potential to impede progress towards gender equality. Stakeholders, including the private sector, policymakers, users, and developers, must grasp the opportunities and challenges posed by AI and take proactive steps to address its deficiencies.
