Monday, December 16, 2024

Which AI skills benefit women at work?

Must read

Much of the discussion around artificial intelligence at work centres on how the technology may make us redundant. But in reality, AI is more likely to become a key tool for workers – and competence in using it is already fast-becoming a sought-after skill for employers.

The problem is that more men than women are frequently using AI at work. While 54% of men now use tools like ChatGPT in their professional or personal lives, this figure falls to just 35% of women.

This AI gender gap not only amplifies existing workplace gender biases like the pay gap, but entrenches discrimination within AI systems. After all, who creates AI tools – and contributes to the data it uses – has the potential to reinforce inequalities instead of breaking them down.

One key reason why AI adoption is slower among women is because of a long-standing lack of women in science and technology.

“Historically, women have been underrepresented in STEM [science, technology, engineering and mathematics] fields, which has created a cyclical effect where fewer female role models and mentors are available in AI-related positions,” says University College London Professor Rose Luckin, who researches how different technologies like AI and machine learning can be used in education.

“Additionally, many AI tools and systems have been developed predominantly by male teams, sometimes leading to interfaces and applications that don’t fully consider diverse perspectives and needs,” Luckin adds.

Additionally, says Dr Fabian Stephany, assistant professor for AI and work at the University of Oxford, women disproportionately shoulder caregiving responsibilities.

“This reduces the time and resources they can dedicate to acquiring AI skills or advancing in AI-intensive careers,” he explains. “Workplace cultures and hiring practices in male-dominated industries often deter women from pursuing AI-related roles.”

Discrimination is already a big problem in AI. When researchers at the Berkeley Haas Center for Equity, Gender and Leadership analysed 133 AI systems across different industries, they found early half of them (44%) showed gender bias. A further 25% of them exhibited racial bias.

In a rapidly advancing AI industry, the lack of gender perspectives, data and decision-making will only worsen the problem. The uptake of flawed AI-powered technology in jobs, education, healthcare and more could lead to biased decisions, discrimination and ultimately, poor services.

For example, artificial intelligence models play a key role in medical diagnoses. But studies have found that these models don’t always work well across all demographic groups – and are more prone to inaccuracies when diagnosing women and people of colour.

Latest article