Artificial Intelligence (AI) bias has come to the fore in recent years, with AI increasingly being used to shape the world around us. The likes of Siri and Alexa, Apple and Amazon's virtual assistants respectively - have become part of many of our everyday lives - integrated into our now very ordinary household objects. These are of course probably the most famous anthropomorphic examples, but AI functions in many sophisticated ways across industries in global corporate software and governments platforms alike. AI leverages computers to mimic the problem-solving and decision-making capabilities of humans and the business benefits are well established with automation and efficiency being great for the bottom line.

Often, despite the best efforts, human biases have made their way into these artificial intelligence systems. In 2018 Amazon scrapped their AI recruiting tool when they discovered that it favoured men for technical jobs. Other documented examples include racial bias in facial recognition technology used by law enforcement in the US and a prejudiced recidivism assessment tool used by the US courts in legal proceedings.

Machine bias is human bias

Bias in AI results when systematic errors in an algorithm create unfair outcomes, such as privileging one arbitrary group of users over another. This bias occurs for many reasons, it is generally an unintended consequence of the human decision-making process that took place when the algorithm was developed. This includes decisions relating to how data is collected, coded or used to train the algorithm.

'Responsible AI' or 'Ethical AI' are now part of many technology firm's training programmes in an effort to drive awareness and combat bias internally. With the advances and proliferation in machine learning, it’s good to remember that with great power comes great responsibility.