Q&A: What AI Means for Digital Media

Online journalism is having another existential crisis. Once-flashy digital-native outlets are experiencing a reckoning: BuzzFeed shut down its news division, Vice has filed for bankruptcy, and Insider is facing strikes after announcing its first-ever mass layoffs.

These outlets were part of a huge shift for publishers, one that made journalism just a subsection of what they produced: almighty content. Their downfall feels like the end of an era — if not a total Ragnarök, given these outlets are floundering right at the moment where generative AI tools like Chat-GPT are starting to become available and promising to table-flip the industry.

We spoke to Felix Simon, a researcher at the Oxford Internet Institute who studies the digital journalism industry and AI, to find out where digital journalism is headed in a post-Chat-GPT world, and whether the generative AI craze holds some of the same pitfalls for journalists that social media once did.

This interview has been edited for length and clarity.

So these digital-born outlets, why are a bunch of them experiencing a crisis point?

I think it’s just a combination of many structural developments, and the sort of root cause is they never had a sustainable business model.

These digital-born outlets, the vast majority of them were sort of predicated on growing through platforms. They made themselves more or less completely dependent on large social media platforms like Facebook, like Twitter, and sort of pursued growth at all costs. And the growth was then used as an argument to go to investors and say, “Hey, we are growing, we can reach international audiences in the millions and millions and millions. And that’s a reason why you should give us money, because at some beautiful point down the line, we will be able to turn this into profitable business.”

That was the promise. The problem is the promise strongly depended on platforms playing along, and they didn’t because platform companies have their own incentives. They’re not necessarily aligned with content-producers at scale, they’re not necessarily aligned with journalistic outlets. So pursuing growth on these classic platforms as the only business model really was never going to work. And now with some general downturn, the writing was on the wall.

I want to talk to you a little bit more about publishers’ relationship with platforms. Right now we’ve got generative AI causing a big scene. And a lot of the same companies that were the ones that publishers had this sort of resentful relationship with are the same companies that say that they’re going to be the ones to really commercialize generative AI. Do you think that could make the relationship with publishers more hostile?

Yes. In some ways the relationship between platforms and publishers has always been difficult and uneasy. They are frenemies in many ways.

With AI, there is some risk that it will further strain the relationship, mostly because it’s increasingly a technology that can produce content. And especially in the realm of how people get to see information, I think that could cause further trouble in the relationship between platforms and publishers. Mainly because if you take something like Bing or Google’s latest search engine you have a situation where as a user you search for something and you get presented with an answer that’s readable and contains all the important information. Then you probably don’t have an incentive to actually go to the original sources where that came from.

Google, and Bing, as well, as they’re saying: “Don’t worry, we’ll make sure that it will link to the original sources.” But even there’s a worry that will lead to a situation where people will actually only get to see search results with the short answer, and that will be enough for them, they won’t actually click through through the publisher. Of course that means less traffic and, by extension, fewer views-slash-impressions on the advertising, and also potentially it could make it harder for [publishers] to get subscribers.

Some outlets are already talking about how they want to incorporate Chat-GPT and generative AI into the writing process, and I saw a tweet the other day that said “pivoting to AI is the new pivoting to video” [for our readers’ edification: this is when news outlets prioritized video because it performed well on Facebook for a period of time] I wanted to get your take on how fair a comparison that is.

I don’t think it’s entirely that, but I think it encapsulates some of the problems. Part of what we see around generative AI is clearly hype, fostered and furthered by these companies who are developing it. And of course, they have an incentive to keep that alive, because it will mean business. Even the negative coverage, all this stuff around “Oh, we are developing super-intelligence,” even that helps their case, because it’s founded on the assumption that the technology they have is really that earth shattering, which may or may not be true. We’ve seen some cases where it does kind of miraculous things and the functionality is quite impressive. In other cases, we’ve seen it hilariously fail.

For publishers there’s this risk at the moment when everyone is pivoting to AI that it’s not done with a consideration for what publishers actually need. In many ways the needs of publishers haven’t really changed, they still need to gather and process information, they have to reach audiences with that information, and they have to make money somehow.

And it’s not entirely clear that in all of these cases AI will make a big difference. It can help in some cases, if you use it wisely and if you find use cases for it where it makes sense, but it’s not the be all and end all.

In some quarters of the news industry there’s this expectation: “We should definitely pivot to that. And it’s going to be the future and all will be well.” And that’s a bit problematic, because ultimately it’s dependent to a large degree on these companies. What they do with this, it’s proprietary, you don’t have access to the underlying models. They’re very unreliable in some cases, we don’t quite know why they are unreliable because we can’t really interrogate them. And if you become too dependent on this as a publisher, of course it can mean something similar as in the pivot-to-video case. Because that was basically something in the control of in this case Facebook, and once Facebook decided it had had enough and moved on, publishers who had built everything on this pivot-to-video were left out in the rain. And with generative AI it could be similar. It doesn’t have to be, but there is this risk.

We started this conversation by talking about this generation of companies that sprung up that have now hit a rough patch. We seem to have a new mini-generation coming through now, I’m thinking about Semafor and The Messenger. Is there anything that you’ve seen in new news outlets coming out that looks like maybe they are avoiding some of the landmines that BuzzFeed and Vice have stepped on?

Yes, [outlets] like The Daily Maverick in South Africa, which basically has done the really hard work of engaging with audiences, building a trusted base of people who will subscribe who will support them in a very difficult climate. Same as Rappler, actually, in the Philippines.

Outlets who’ve made it their task of really walking the walk and really putting meaning to the words “engaging with audiences,” with communities, making sure that they produce journalism that is locally and nationally relevant. And finding business models made of combinations of things. That could be philanthropy combined with some advertising, combined with subscription models, combined with events organization, which altogether sort of carry them and allow them to do journalistic work. I think in many ways those are much more interesting business models, at least to me, than outlets which are currently completely funded by venture capitalists.