A new study has revealed an AI bias against women in leadership. 

Researchers have conducted a thorough analysis of AI-generated material that evaluates the qualities of ‘good’ and ‘bad’ leaders. The results were stark, indicating a consistent bias against women in leadership roles.

In the investigation, men consistently emerged as strong, courageous, and competent leaders, while women were often depicted as emotional and ineffective. 

“By exploring how AI tools generate responses to questions about good and bad leadership, we were able to show that AI reflects and perpetuates known gender biases,” said the University of Tasmania's Dr Toby Newstead.

One glaring issue was the complete omission of women leaders from the initial data generated by the AI, with no examples provided until specifically prompted. 

Even more concerning was that when asked to generate content about women in leadership, the AI disproportionately portrayed them as bad leaders, falsely implying that women were more likely to be ineffective leaders than men.

Generative AI technologies, which rely on machine learning processes to generate content, are trained using extensive data from the internet and human input aimed at reducing biassed information. 

Despite these efforts, the study shows there is a need for further scrutiny and investigation into AI tools as they become integral parts of daily life.

“Biases in AI models have far-reaching implications beyond just shaping the future of leadership,” said researcher Dr Bronwyn Eager.

“With the rapid adoption of AI across all sectors, we must ensure that potentially harmful biases relating to gender, race, ethnicity, age, disability, and sexuality aren’t preserved.

“Our research highlights the need for ongoing monitoring and evaluation of AI-generated content to ensure that it is not perpetuating harmful biases. We hope that our research will contribute to a broader conversation about the responsible use of AI in the workplace.”

The full study is accessible here.