Is AI biased? "We start by confirming that indeed AI is biased," said Josie Christodoulou (
@JosieChristodou), Commissioner for Gender Equality (
@CommissionerGE) of the Republic of Cyprus, in the fireside discussion "AI and Equal Participation: Who Shapes the Digital State?", chaired by Andreas Demetriou, founding member and Legal Advisor of AI³, and Rodoula Papalambrianou, Scientific Coordinator at Hope for Children (
@HFC_uncrc).
Commissioner Christodoulou explained that models learn from the data they are given. "If you give a model a thousand books where in all 1,000 examples the doctor is a man and the nurse is a woman," she said, it will assume the next doctor is a man. "It perpetuates the bias." Breaking the cycle starts in schools, giving girls who want to pursue science and technology real opportunities, and continues into the workplace, where many women with the right skills still end up in research or academia rather than industry. The question, she argued, is not only whether women and minorities fit into the system, but whether we really want to change it. "Technology must be designed to serve women. The decision-making must include them."
Ms Papalambrianou turned to children and the ways AI is being used for violence, including deepfake technology to create sexualised images and spread them through social media. Rather than choosing between restricting access and open access, she argued, the answer is educating children. Many fully trust technology without questioning it, and some use AI as a trusted advisor, "an available adult" always there to listen. They need to be empowered to build a critical stance, with a unified approach to digital literacy for all children and special attention to the most vulnerable. "It will not be solved by an algorithm," she said. "A human must be involved."