Machine learning expert warns of 'toxic' AI future

Imagine a future where you are regularly stopped and searched by the police, based simply on bad information fed into a computer. That is the fear of one authority on the subject, who is concerned that human biases and errors are being programmed into machine learning systems. Studies have already shown that AI experiments, including predictive policing, demonstrate exaggerated versions of human biases when put into practice. Experts are worried that this could lead to a'toxic' future where machines make bad decisions on our behalf, unless something is done now to prevent it. Imagine a future where you are regularly stopped and searched by the police, based simply on bad information fed into a computer.