Are the algorithms that exist in our daily lives getting so smart that they know us better than our parents or our spouses? How do we balance the convenience and efficiency of this technology with privacy and consumer protections?
Sandra Matz is a professor at Columbia Business School and the director of the Center for Advanced Technology and Human Performance. Her book, Mindmasters: The Data-Driven Science of Predicting and Changing Human Behavior examines the link between algorithms and psychology.
Sandra and Greg chat about the bright and dark sides of psychological targeting, its applications in marketing, politics, and mental health, as well as the ethical considerations and future implications of using algorithms for personalized interactions.
*unSILOed Podcast is produced by University FM.*
Episode Quotes:
Even the smartest algorithms slip up
24:01: When we talk about these algorithms, and I'm guilty of that myself, it always seems like, well, yeah, if they can predict your personality with high accuracy, that makes sense, but it still makes mistakes, right? So, accuracy is always captured at the average level. So, on average, we kind of get it right most of the time. But that still means that, at the individual level, we make a lot of mistakes. And those mistakes can be costly for the individual, right? 'Cause now you are seeing stuff that is completely irrelevant. Also costly for companies, 'cause now you are optimizing for something that's not actually true. So, I think if you can really think about application—I think the more you can turn this into a two-way street and conversation, the same way that this works in an offline world, right? If you kind of suddenly start talking to me about topics that I care nothing about, you're going to get that feedback, 'cause either I'm not going to see you again, or I just tell you we just talk about something else. And companies oftentimes don't get that because they don't allow users to interact with some of the predictions that they make. And I think it's a mistake, not just from an ethical point of view, but even from a kind of service, convenience, product point of view.
Are algorithms making us boring?
11:09: There's something nice about having these algorithms understand what we want, but I also do think that there's the risk of us just becoming really boring.
The trouble with signing away our data
49:29: The way that we typically sign away data is, we consent, but not because we understand it. And I think some of it is just that technology moves so fast that just keeping up with technology is almost impossible. So I think about this 24/7, and I have a hard time, and you also have to have this understanding of — not just in the here and now — like, a fully rational person would say, "Here's all the benefits, and here's all the downsides." And now I kind of make this rational decision that kind of maximizes utility. But we don't understand the downsides.
Show Links:
Recommended Resources:
Guest Profile:
Her Work:
