JOE HALPERN: My name is Joe Halpern. I am the Joseph C. Ford Professor of Engineering at Cornell University, and I'm also the chair of the computer science section of AAAS, which is why I'm the organizer of and moderator for a panel on AI and ethics that will be held at AAAS. Let me give you briefly some of the background that inspired us to have this panel.
So AI and machine learning have been around since the '50s. And back in the '80s, there were some very successful AI expert systems that were designed to simulate experts. Now, they were all rule-based, so that at least in principle, sometimes their conclusions were surprising, but you could sort of always look at them and understand why they were doing what they were doing because the rules were written by people, and certainly interpretable.
Now, what's happened in the last decade or so, we have massive computing power and enormous swaths of data. Think Google, Microsoft. And we have machine learning algorithms that aren't rule-based, but instead look at the data, and try to learn from the data, and are working incredibly well.
But the trouble is that now, we can't always understand why they make the decisions they do. So if we have a machine learning system that's deciding who gets the loan, or who gets the job, or who gets to get paroled, we might not understand why it's making the decision, even though it might, in fact, be a reasonable decision.
And it leads to a whole host of ethical issues. Are the decisions fair? What does fair even mean? Is it OK to have the machines make the decisions? Is it OK to have autonomous vehicles have self-driving cars?
And if that's OK, is it OK to have autonomous weapons that decide for themselves where they're going to aim and who they're going to target? The AI community is aware of these issues, and the point of the panel is to bring out the issues and make the community beyond the AI community aware of what we're doing. And it really should be a societal discussion.
We've received your request
You will be notified by email when the transcript and captions are available. The process may take up to 5 business days. Please contact firstname.lastname@example.org if you have any questions about this request.
Joseph Halpern, Joseph C. Ford professor of Engineering at Cornell University, studies Artificial Intelligence and ethics and will be moderating a panel on ethical questions in AI at the American Association for the Advancement of Science 2020 Annual Meeting in Seattle.