AI should be regulated as machines can make bad decisions impacting lives. But companies are so secretive about their algorithms. We have seen how Google and Uber went on the head to head with their famous lawsuit this year. What I wonder is, is it possible to have a third party acting as a supervisory body to audit these algorithms? In the UK, a company doesn’t have to disclose a trade secret, so how will anyone regulate these AI algorithms for correctness or safety? I read that in 2018 EU would have new laws where AI can be challenged, so we’ll have to wait and see how they implement it.
Alan Winfield, professor of robot ethics at the University of the West of England, is heading up a project to develop industry standards for AIs that aims to make them transparent and so more accountable. “A watchdog is a very good proposal,” he said. “This is not a future problem, it’s here and now.” But he agreed that tech firms might struggle to explain their AI’s decisions. Algorithms, especially those based on deep learning techniques, can be so opaque that it is practically impossible to explain how they reach decisions (Sample, 2017).
Sample, I. (2017) AI watchdog needed to regulate automated decision-making, say experts. Available at: https://www.theguardian.com/technology/2017/jan/27/ai-artificial-intelligence-watchdog-needed-to-prevent-discriminatory-automated-decisions (Accessed: 19 June 2017)