"Technology is like a mirror of society, simply reflecting what we as individuals bring to the table"

From discriminatory algorithms to racist soap dispensers: artificial intelligence (AI) is not as neutral as it might seem at first glance. Our AI expert explains the reasons behind this in the taz.

When digital applications discriminate, it can also be due to the people involved in their development. A homogeneous group may unknowingly design products that do not work equally well for everyone because their own unconscious thought patterns and stereotypes are transferred into the product, or because they fail to consider the diversity of all end users. In Germany, men are primarily involved in IT development. In 2018, they accounted for 83 percent of employees in the IT sector. Existing inequalities from the analog world are thus effectively transferred into the digital space.

This can lead to situations where women are disadvantaged by voice processing in online conversations, for example. Parts of their speech are sometimes not transmitted because higher frequencies are filtered out by the programs. The result: contributions from women are often perceived as significantly less convincing.

But gender is not the only characteristic that is not sufficiently taken into account in the development of IT products. Many applications based on artificial intelligence (AI) are also primarily designed for the needs of white people. EAF expert Lisa Hanstein explains it this way:

"If we only train facial recognition modules with images of white men, those modules will naturally be very good at recognizing white men, but not Black women, for example."

However, the "Decoding Antisemitism" project shows that AI can also be used to combat existing discrimination. According to Hanstein, it was developed to identify antisemitic comments on social media so that they can be deleted more quickly and effectively.

The AI expert's conclusion: A consciousness for societal discrimination mechanisms is slowly and gradually taking hold in the IT industry as well.

Read the full article here.