Applying ethics to AI: what to how By Caryn Tan How is it that algorithmic bias is one of the hottest topics in artificial intelligence right now? Isn’t the point of using machines to make fair decisions removed from human bias? Aren’t machines neutral after all? Initial thinking believes that algorithmic decision-making results are not only more accurate but also fairer than more traditional methods. This is because the machines are not prejudiced by racial, gender or socioeconomic biases from our past. Despite machines being neutral, the machine is learning from data which captures our messy past as well as prejudices and bias from history. This means a neutral machine will pick up on these bias patterns and potentially result in unjustifiably biased predictions for future decisions. An assessment of parameters used to define machine learning algorithms and the data sets used to train these algorithms is key to ensure that we are not building machines and autonomous systems which are making unfair outcomes. Not only should we be assessing this for a qualitative perspective, it is important to start applying quantitative methods coming out of academia and research labs to the difficult question of fairness in algorithms. The imperative for businesses to think about this is to maintain trustworthiness with their stakeholders. In this age of information, consumers are becoming more savvy and choosing certain products and services by assessing for factors beyond price into considerations such as ethics, social responsibility and sustainability. As companies gain more power through having more data and sophisticated ways of analysing and gathering information about their stakeholders, being a trustworthy organisation which can avert risks of unfair outcomes will become the next currency of competitive advantage. If you liked what you just read, why not sign up to our weekly newsletter? Get all the content straight to your inbox here.