Matz had just been correcting the record on Cambridge Analytica: it was not a breach, Facebook handed the data over by design. Laurie Segall asks whether the scandal now looks like a warm-up for AI. Matz says the underlying principles have not changed, only the scale, since a model no longer needs someone to assemble the dataset and train on it first.
I mean, I think he's actually right in that they did restrict access. And that's something that I always felt was actually not necessarily accurately reported about Cambridge Analytica because it wasn't a data breach. It wasn't that someone went in and stole the data. Is Facebook at the time was willing to give the data. Back in the day, it was the moment that I shared my data as a Facebook user, the developers got access to my entire network of friends and all of their data. So what Alex Kogan, the guy that he mentions in his response, what they did was actually completely in line with Facebook's policies. They were not allowed to sell the data. So I think that's the mistake that they made. But the data was actually open to developers to grab. And to me, yeah, they did restrict the access after. But again, this is something that when you build technology and you move fast and you break things, well, maybe that's an oversight that happened because they were not responsible enough to start with.
It's almost like more alarming that this was just pay. And actually, and it's interesting to go back to Cambridge Analytica because I think a lot of people could say, oh, this is like water under the bridge. This was a scandal a long time ago. But I think it's important, even from what we both talk about, to look back at that and now look at it through the lens of artificial intelligence because to some degree, it feels like to me that was just the warm-up with what we're doing. We're now about to see with AI and all of the complexities and all of the questions about data that are now front and center. And we are in this kind of new era of move fast and break things, only like it feels like the stakes are even higher. And that could just be me putting my own point of view on it. But I'm curious how you feel about looking back at that and where we sit now with artificial intelligence and the questions that we're now beginning to ask about it.
I think Cambridge Analytica looks cute in hindsight now with AI. I mean, the one thing I remember when I published the book, I was like, oh my God, this is going to be outdated so quickly because technology moves so fast. But at the end of the day, the main principles remain the same. And if anything, I think AI is just accelerating it. And so the challenge that you have with AI that's different today than back in even 2016 is simply scale. So back in 2016, Alex Kogan had to collect the data set. Then they had to train models. So they had to combine psychology with data points, kind of train their proprietary model, make the predictions. Then a human had to go in and say, okay, based on these profiles, we're going to come up with this type of messaging. We're going to have segments. Now, all of this is completely obsolete. But I can take all of your data, it doesn't matter which format it comes in, I can pop it into ChatGPT, Claude, Gemini, you name it, and just say, well, tell me who do you think Laurie is? Give me a sense of what her personality is, her values. And by the way, while you're on it, also just create me a bunch of messages that you think would resonate with Laurie in a certain context. And it's incredible at doing that. So it's no, you don't really need humans in the loop anymore. You don't need proprietary models. And you don't need anyone to come up with the messaging because AI can do it so much better than humans can.
It's one thing I heard some folks talking about, kind of like, you know, in the corners of whatever, like the rise of wearables and AI wearables, like these rings that can listen. Like we're kind of, we're beginning, people are going to start recording everything, whether or not we even realize it. And how could that be used for political targeting and all of these ways? So it certainly feels like we're just like, I think you're right. I think Cambridge Analytica feels cute compared to what we're about to look at.
And it's a great example, because I think what people associate with Cambridge Analytica is, well, I'm just not using social media anymore. It's a comment that I hear almost every time that I give a talk somewhere. It's like someone feels very proud that they're not using social media anymore and they just feel safe.
And