How Do We Keep Bias Out of AI Programmes?