How to Make Fair Algorithms ~ Talk

I recently went to a talk at The Royal Society in London as a part of their You And AI series entitled 'You and AI - The emerging theory of algorithmic fairness: the challenges to making machines play fair'. 

Speaking was theoretical computer scientist Cynthia Dwork, a Gordon Mckay Professor of Computer Science at the Harvard Paulson School of Engineering, Radcliffe Alumnae Professor at the Radcliffe Institute for Advanced Study and Affiliated Faculty at Harvard Law School.

After analysing how to navigate "fair challenges" and the mathematics behind 'fair algorithms', Professor Dwork revealed why simply hiding sensitive information about people from the algorithm can in-turn end up ignoring the issue of bias-ism without adequately addressing it. Professor Dwork also pointed out that; 

"A culturally aware algorithm may be a more accurate algorithm."

​The Full Talk

The issues of bia-ism as a result of a non-diverse work place in AI and technology and/or non-diverse input into algorithmic compositions are clear. However on our way to defining fairness and whilst analysing algorithms and the method of their design, much needed conducive realization becomes apparent. The realization of our own biasness and how it may be impacting the world around us and how we go about our lives. The realization of the bias community that we may be living in and how implementing this into something as powerful and societal impacting as AI is dangerous for humanity if we wish to continue forward and develop fair algorithms.

Unfair and bias algorithms in AI will not only reinforce the issues of direct or indirect discrimination which we have in society, but in its mass impact on society, is in danger of amplifying them.

Another important point which Professor Dwork highlighted was that the methods to test fairness in algorithms should be monitored consistently. During the questions and answers section at the end of the talk, I asked the Professor who she felt was best suited in the role of testing these methods; the robotic engineers, ethical consultants, the government..? To which she replied;

"All of the above. And in public."

I liked that answer...