For a long time lauded as the way to automate large parts of the industry, AI is, according to Lindsay Silver, VP of Product at Condé Nast, finally coming of age – and AI 2.0 will have plenty to offer beyond simply taking tasks away from the editors. In an interview with Jon Watkins of FIPP, Lindsay Silver talks about how AI will transform the media world.
Jon Watkins: Does it feel like AI is finally getting real for our industry after a long period of speculation about what it will deliver?
Lindsay Silver: It does feel like we’re almost at AI 2.0 in terms of what people are doing with content. At 1.0 it would be all about recommendations and personalization. But how do you build models and machine learning that captures context or captures a person, and actually reacts to that? Obviously, the early stuff was pioneered by Facebook and Google, to some extent, and is now pretty mainstream. And companies like Condé Nast, and almost every other first-tier media company, are using some personalization. But that’s old news now. People are looking for that jump to 2.0 and that means a lot of different things depending on who you are. For Condé Nast we’re trying to figure out what differentiates us. And I think the really interesting thing there is whether AI can capture editorial intent? Can we have an AI that actually has an opinion? That’s something that I think will be extremely important to companies like Condé Nast, if we’re going to actually use it in an advanced way. So that’s what we’re working on – how do we create machine learning models that react editorially or take an opinion about what’s shown? And what happens next?
JW: Where will the big wins come and how will it change the way we work?
LS: I think there are a lot of different channels where there are big wins to be had. If I look at what an editor’s job is and really boil it down to the atomic units, I want an editor to be creating an idea – developing it through images, or developing it through words, or developing events, or video or other alternatives, and I want them to be doing that in its purest form. I don’t want them to be spending time building hyperlinks, auto linking products, uploading stories. That’s what AI can do. One of the things we’ve done for a while now, which I think is really interesting, is we’ve done a lot of auto linking of content. We’ll look at a piece of content. And using a machine learning model plus a set of business rules, identify the things we should be linking to that maybe an editor didn’t think of. The machine can identify where the best place is to link to, what would actually get the highest click-through rate or a certain bounce rate. That kind of thing is interesting, especially when you introduce editorial content into it. What is it that this editor would want me to see when I link off to that? We do some fairly basic but personalized versions of that across our sites and we’re continuing to evolve that and learn and be better at that.
JW: How might AI disrupt our market in terms of the types of players we have – and for publishers versus the platforms, for example?
LS: My feeling is that that the platform businesses and even companies like Condé Nast, who are successful in building a solid and stable platform, will benefit from content in its purest form. Like Netflix has done for video, I think it’s likely that there will be a player that will surface that will be extremely good at providing a platform for that. And I think at the other end of the spectrum, like HBO or some of the niche video players have done, I think there will be content producers who will have enough power over their corner of the content market, that they will manage to sort of create the products through that. I’m not somebody who believes that AI will be writing full articles in the New York Times, but it will turn the focus on to quality experiences through content and it will help you deliver it.
JW: What’s driving the rise of AI? Is it demand from audiences for a better experience of the drive of companies to find that competitive edge?
LS: I think for us, it’s advantage in the market and it’s finding a point of differentiation – not in terms of differentiating against the New York Times, for example, but differentiating against the platform players. That’s why we’re so focused on editorial intent. We’ve grown our data team significantly. Over the last few years, we’ve invested a huge amount in machine learning and data science. I’m not going to compete with Google or Facebook. But what I can do that they can’t do is focus in on what my editors actually need. And what I really like – and we came up with this concept a long time ago – is that Google is always going to be reactionary. It can only act after something is published. My advantage is that I know what’s happening. The value of Condé Nast becomes that from the point that we know an idea exists until the point that it’s published, we have it and no one else does, right? So, we’re investing more and more in how we can isolate that? How do we empower our teams to understand what’s important and get that out?
JW: How do you manage that investment in AI and your development of it internally?
LS: We haven’t built dedicated teams around innovation or anything like that. What we have said is that AI is a vertical. And it’s like everything else, it’s 90% kind of drive to work and 10% the magic. We started by building a large data team – we have about 100 people on our data team just focused on building that platform. And we separate that into an audience platform and a content platform. The more we can derive from our content, or decorate our content with knowledge that we have, the better the models will be at a higher level. And we have a data set, we have a data science team that sits on top of that to actually build the models and to create these things. What we’ve done that I think is similar to Facebook is we’ve said to every product feature team, you have a mandate to make these processes better using AI. The best of our teams are pushing that forward, and everyone else is following. Things like auto linking, personalized recommendations, some of the work we’ve done inside of our content management system – we’re doing a really neat project around creating all text for images, using AI – is being driven by those guys. And, you know, like everybody else, we always wish we were doing a lot more, but we’re doing a decent amount. I think we’re definitely on par with the best in the media industry.
The article was first published on FIPP.