Sandjai Bhulai, professor of Business Analytics at VU, opened the symposium with a striking demonstration. He asked AI software to generate portraits of doctors and board chairs. The result: exclusively white, middle-aged men. For financial advisors, ten men in ties appeared, while the only woman was depicted with bare arms and blonde wavy hair. Bhulai emphasised that these images are not the result of fundamental errors in algorithms, but of biases in training data.

The professor, who helps companies with data-driven decision-making, presented concrete examples from healthcare. In pain recognition studies, algorithms are shaped by historical data from periods when pain in women was taken less seriously. This leads to women sometimes receiving less pain relief than necessary. Oxygen measurements also pose a problem: these are usually performed on white skin, meaning the resulting data are not representative for people with darker skin.

Bhulai stated that algorithms hold up a mirror to society. Research shows that women see fewer online tech advertisements and that when applications determine who is entitled to salary increases or promotions, women score significantly lower than men. Following the presentation, a roundtable discussion took place with VU researchers about the practical consequences of these technological biases. The symposium notably attracted not only the traditionally large group of female visitors, but this time also men in large numbers.