Author: Frank Rudzicz
Publisher: MaRS Discovery District
Publication Year: 2019
Summary: In the following video, Frank Rudzicz, an artificial intelligence (AI) researcher, lays out the 3 big ethical concerns in AI. The first one is what we actually use AI for, when we develop models we create it for a distinct purpose, however that can get entangled. We may use video tracking with people (in healthcare) to keep track of recovery from an injury but the same technology can be used in a malicious way such as a government using it to track their citizens. We need to understand what are the potential outcomes that we do not necessarily expect. The second concern is who has access to AI, as we progress, AI has to run on faster and more expensive machines. Big international companies are usually the only ones who can afford this which means fewer and fewer people can control AI. The third concern is that AI does not think as humans do nor does it share our values. The risk is that AI will do exactly what we tell it to do, this happens when we tell AI what we want but define it in a vague manner. The consequence of this is bias; the AI may adopt bias from the data it was trained on. We need to prompt an open discussion on what AI can and cannot do and how we can manipulate things to make sure it can be used to the benefit of as many people as possible.