The new wisdom: According to the author Jeffrey Norris, source UC Berkeley, compiled by the new wisdom exclusive compilation, refused to reprint without permission!
How do we make artificial intelligence and robots live in harmony with us as we become more and more globally automated? The answer from the University of California, Berkeley, is: They will set up a new artificial intelligence center this week to explore how to make robots and humans coexist in harmony. It is reported that the director of the center will be served by AI expert Stuart Russell, who is a professor of electrical engineering and computer science at Berkeley. He holds the title of Smith-Zadeh professor of engineering and is also a collaborator of Artificial Intelligence: A Modern Approach. This book is a benchmark in the field of artificial intelligence. It advocates the integration of human values ​​into AI design.
It is understood that this center is a master and has strong scientific research strengths, including Pieter Abbeel from Berkeley, Anca Dragan, and cognitive scientist Tom Griffiths, computer scientists Bart Selman and Joseph Halpern from Cornell University, AI experts from the University of Michigan Michael Wellman and Satinder Singh Baveja. Russell said that in the future he hopes to have experts from economics, philosophy, and other social sciences. The main goal of this research center is to ensure that the AI ​​system benefits human society. In addition, the center received 5.5 million startup funds from the Open Philanthropy Project and also received research grants from the Leverhulme Trust and the Future of Life Institute.
Russell's attitude as an expert quickly dismantled the threat that people may have on the sci-fi blockbusters that are aware of the evil robots. He believes that the robots we now research and develop in the fields of AI, robotics, control theory, and operational research are completely human. If you let the home robot clean up the bathtub, it may be like the cat in "The Cat with a Hat", not knowing that a clean white dress is worth more than a clean bathtub, and thus - wipe it with her mother's white dress Bathtub. In other words, at present, the design of AI does not agree with human values. This research center will use the most advanced AI system to provide necessary services for tens of billions of people. He also stated that “the AI ​​system must be controlled and there must be clear restrictions on behavior, although they may eventually have some capabilities that humans cannot reach.â€
At present, Russell and other experts are developing reverse reinforcement learning. In this process, robots learn human values ​​by observing human behavior. For example, by observing the pain of people getting up in the morning and trying hard to get a cup of latte after getting up, the robot will recognize the significance of the existence of a morning cup of coffee for people.
"It is absolutely undesirable to directly embed certain values ​​in the algorithm," Russell said. "Robots must learn from human values ​​by observing. Reading is an important way, in addition to observation. All text provided to robots is About how humans do things and what others think.â€
However, Russel and his colleagues all know that they still have a long way to go in the AI ​​field. "People have different kinds of values, so if you want to achieve all values, you can't do it," he admitted. "Different values ​​make robots try to understand what we want them to do, and learn to mediate between individuals."
In addition, Russell recently wrote an article called "Do They Make Us Better?" His conclusion in the text is: "AI will agree with human values, and this process will give us a better understanding of what we are doing. In addition, we should probably recognize that we are from other cultural backgrounds. There are more similarities in people."
Via UC Berkeley
Xinzhi made Lei Feng network (search "Lei Feng network" public number attention) (search "Lei Feng network" public number attention) (search "Lei Feng network" public number attention) under the focus of the robot's public number, we are concerned about the status of the robot and The future, as well as related industries associated with robotics. Interested friends can add micro signal AIRobotics, or directly scan the QR code below to follow!
