The trouble with artificial intelligence, as Patrick McDaniel sees it, is that most people don’t really understand it — either its strengths, its weaknesses, or its potential. Thus, much of the conversation about AI’s implications considers the wrong things.
“We fundamentally don’t understand what the market impact of AI are going to be,” he says.
McDaniel is the UW’s Tsu-Ming Shih Professor of Computer Sciences, and he’s the associate dean for research in the new College of Computing and AI. He studies computer security and leads the National Center for Trustworthy Machine Learning, and over the last decade, he’s been on the forefront of the AI revolution. On May 19, he’ll join host Mike Knetter and economics professor Kim Ruhl on the UW Now Live for a discussion about the economy and the effects that AI likely will — and won’t — have.
My chief Area of Research Is:
I’ve led various national and international organizations around cybersecurity. I work principally as a researcher in issues related to cybersecurity as they relate to mobile systems, elections, and, most recently, AI. I’m going to be the leading administrator for AI research at the University of Wisconsin.
Tonight on the UW Now Live, I’ll Talk About:
I’m going to give three propositions — a triad of basic misunderstandings people have about AI and its impacts. The first one is the human capital fallacy. There have been a lot of stories about companies laying off technologists and saying it’s because of AI. But there’s something called the Jevons Paradox, which is an economic theory that basically says, in times of major change, you don’t need fewer people, you need more. Judith Faulker of Epic [Systems] fame had a great, great quote, which says, “People who think that AI is going to take over everything don’t really understand the complexity of human work.”
The second misunderstanding is about how AI works as a technology. The reality is that AI is an inherently open technology. It’s not a technology that one company can develop and keep to themselves. The AI wars are just not really relevant, because no software company can really win. Open models will perform well enough that no software company will be able to establish or maintain dominance.
And third, and most important, there’s a fundamental miscalculation about why the public is rejecting AI. In Silicon Valley, they think that people reject AI because they’re afraid for their jobs. But really, the reason it’s being rejected is because people don’t trust it. Big tech has continually provided these soaring promises, but they just don’t deliver. AI isn’t going to work on behalf of users; it will work in a way that maximizes [those companies’] profit or their ends and not yours. So people mistrust it, and they’re right to.
If Viewers Remember Just One Thing, It Should Be:
With AI, we are in an undiscovered country. We really need to not listen to the evangelists or the naysayers. We need to be thoughtful consumers.
To Get Smart Fast, See:
In general, the people that I have found to be the most credible are economists who are looking at real impacts. Look at good studies, recent studies. They show that all of this magic AI that writes code doesn’t actually provide as much capability or productivity as you think. So stay tuned, watch the numbers, listen to a broad range of people, and stay away from the hype curve.
