Melanie Mitchell, Computing Community Consortium (CCC) Council member and Professor at the Santa Fe Institute and Portland State University, was recently on the Munk Debates podcast, in an episode titled “The Rise of Thinking Machines” with Stuart Russell, Professor of Computer Science at the University of California at Berkeley. The podcast, led by Munk Debate chair and moderator Rudyard Griffiths, explores whether the quest for true AI is one of the great existential risks of our time.
Russell believes that if we keep on our current path, AI has the potential to cause real harm. He said, “We need to build a new foundation for AI, but we don’t know how long that will take and how long we have left.” He used King Midas (the King from Greek mythology who turned everything he touched to gold) as an example. We create AI to have a similar fixed objective, not to turn everything to gold, but instead to maximize clicks on a webpage. However, if we keep focusing on one objective, it tells the AI that the rest of the world doesn’t matter and there are no consequences. Like Midas turning his family and food into gold, Russell argues that keeping on this path and continuing to develop AI with its fixed objective, could result in grave consequences.
Mitchell, on the other hand, disagreed with Russell and this notion that AI is an existential risk. She said, “True AI, superintelligence, is not [a short-term existential risk] since we are far from reaching that stage. The last seven decades show that AI research is so much harder than we thought.” Also, we as humans are far from understanding what intelligence is, since it includes many aspects like common sense. How can we design AI to have intelligence without fully understanding our own? Instead, Mitchell argues that we should be aware of the following risks that AI does pose, AI for surveillance, racial / gender bias embedded in AI systems, and the deployment of AI systems that are not trustworthy. According to her, “The most useful thing we can do is to gain a better understanding of intelligence more generally.“
Mitchell is a member of the CCC task force on Responsible Computing. The task force focuses on issues concerning privacy, ethics and overall responsible practices in computing. Listen to more of the debate in the full podcast here.
