Using AI to solve unconscious bias

“Bias – we all have it – the important thing is recognising it and doing something about it. ”


Stefanie Creff and Brigitte Ricou-Bellan talked about the project they have created called ‘Ellpha’, based on their vision of ‘using artificial intelligence (AI) to create a gender-balanced world, faster.’

The starting point for Ellpha was a sense of dissatisfaction with the unequal share of economic power that women have in the world and a frustration at the slow rate of progress, especially when there is so much data to show that diverse teams perform better.

Could there be a tech solution to such an ingrained society problem? Brigitte and Stefanie believe that AI has a powerful role to play in detecting bias and ‘making visible what is invisible’.

The team has built its first prototype, using data provided by a French children’s publisher to see whether male and female characters are portrayed differently. A favourite example of this type of analysis is the Disney film ‘Frozen’. Although the character Anna appears on the surface to be a new style of independent, strong female character, in reality she is highly reliant on other characters to the point of being incapable of making independent decisions. This is the kind of anomaly that AI can reveal.

AI technology could be used to detect bias in many areas of business, such as corporate communications (e.g. websites, social media profiles) and recruitment (e.g. job descriptions, adverts, feedback given to candidates).

By identifying ‘hotspots’ of bias, the team look to help businesses make their content more balanced. They believe that doing this at scale – in a ‘soft but precise way’ – could start to make a big difference to how organisations make decisions.

The team is looking to develop a prototype in the HR field and need partners who can give them access to internal data. If you are interested in challenging your business to make more balanced decisions, Stefanie and Brigitte would love to hear from you.