By Ren LaForme, managing editor
After decades of navigating technological upheavals, the battered and resilient news industry is realizing it’s time to raise its gloves once more — this time against an insurgent troupe of artificial intelligence companies.
While doomsdayers have long predicted that AI would threaten journalism, it’s finally feeling real this month as reports emerge of a sharp decline in search traffic — a once-reliable source that brings an audience to news sites.
Alex Mahadevan, director of MediaWise and a Poynter faculty member, just returned from a convention of forward-thinkers in the industry. I reached out to ask him what he heard and how apocalyptic their tone was. What followed was a frank but surprisingly optimistic conversation. A version, edited for brevity and clarity, appears below.
Ren LaForme: I saw that you went to the Hacks/Hackers AI x Journalism Summit last week. I wanted to catch up for an urgent reason: I need to know what people are saying about AI right now — as opposed to even a few months ago — because I've had my eyes glued to our traffic numbers, and I'm seeing things that don't look good. But first, I'm curious: Who was there and what did you hear?
Alex Mahadevan: There were journalists, editors, technologists and entrepreneurs who were trying to sell AI tools. Newsrooms of all sizes were represented — The Washington Post, The New York Times, The Associated Press; the big guys. Some great nonprofits: The Baltimore Banner was a local partner — I missed their party, which ended up being a sort of Pulitzer party, the day before — as well The Haitian Times. The vibe was very techy; a lot of hoodies.
It came at a really interesting time, because all of these think pieces and data reports were coming out saying that search traffic was cratering. That cloud was hanging over the whole summit. It was in the backs of all of our minds. One speaker, Courtney Radsch, literally said, “Search traffic is never coming back.”
That said, it was a pretty positive conference, and there were a lot of really interesting AI things that people are doing that will really help journalism. But it felt like we had reached an inflection point. Being at a conference with a bunch of technologists who have lived through the internet age, the social media age, the pivot to video, it feels like we are absolutely in another one of those moments. We don't really know what it means for journalism yet. But the overwhelming theme of the conference was that it's time to figure out how AI is going to help us survive the search apocalypse.
LaForme: I want to focus a little bit on why this matters. And I know you said we can't figure out what it means yet, but there's a pretty clear line we can follow.
I started panicking a little bit about Chartbeat's numbers a few weeks ago when I saw that we were missing double digits of our normal traffic on any given day. And it's almost entirely represented by search. Over the past 20 or so years, search engines have been some of the biggest drivers for traffic to news websites. Even through the era of social media — sending us traffic and then not sending us traffic and making us pivot to video — search, and specifically Google, has been reliable.
And what we're seeing now is that less of that traffic is coming in. Is that an accurate summary of what you heard?
Mahadevan: Yes. Across the board, news organizations are seeing the same thing that you're seeing; that search traffic is down. I won't name names, but anyone I talked to there was feeling it and seeing it.
LaForme: This is a big deal for a couple of reasons. A lot of news organizations still rely on advertising, and if you're losing a quarter, maybe more, of your traffic, you're going to lose a lot of money. But, overall, if fewer people are looking at our work, what are we doing? We publish for people, not for our own amusement. How are we going to get these people back?
You've categorized this as an apocalypse, and it feels like that. I have struggled to convey to people in a way that doesn't sound very “Henny Penny,” “sky is falling.” But I think we're in a pivotal moment.
Mahadevan: Yeah. Aimee Rinehart from the AP pointed out in one of the panels that multiple stories have referred to this as an “extinction-level event” in the news. In another conversation, someone brought up the idea of “Chicken Little” and “the sky is falling,” and whether or not that is actually what is happening. But it is.
There are two interesting studies that might make you feel better right now. One is from Nick Hagar at Northwestern, who studied what people are searching for in ChatGPT. Less than 2% of queries are for news-related topics. So people aren't going and opening a chatbot expecting to find news. They're going there for different things. Honestly, it's a lot of not-safe-for-work stuff. But people are not looking to these tools to replace where they get their news.
The second is a study we did with Ben Toff of the University of Minnesota. We found that far fewer people are actually using these tools than we think. Less than half of the people we surveyed use them on a daily or weekly basis, or have ever even heard of them. A large portion of the population doesn't care about these tools, and actually, half of the people we studied said, “I don't want to get my news from a chatbot.” So that does make me feel better.
These AI technologies are being forced on people. Everyone is downloading ChatGPT, but it's not like they're looking at it as a replacement for news outlets. However, Google is inserting AI summaries into all its searches. So it's not like ChatGPT or Claude or Perplexity is siphoning views. It's Google itself.
Its AI overviews summarize what people ask about, say, the new pope: “Is the new pope a Cubs fan or a White Sox fan?” There’s probably great reporting on that, but users are just going to see the summary — and that's all they care about. The problem with that is that they're not clicking through to news sites — the data shows it.
There are two problems here. One is that these AI tools hallucinate and are known for spitting out false information. If people are going to Google, there's a chance that they're getting bad information if they rely on an AI summary. And, two, it's siphoning traffic away from legitimate news organizations that don't hallucinate.
The other troubling thing is that Nick studied referrals from the chatbots, even though people don't use ChatGPT for news, and their findings are awful for news organizations. If people do start using ChatGPT for news, then we're not even going to pick up that referral traffic. It's just going away.
LaForme: I want to frame my next question in a neutral, non-defensive way, which can be a default when we talk about news outlets and technology changes. I'm thinking back to Craigslist. If you talk to a lot of people in news today, they'll say, “Oh, Craigslist came for us and took all of our classifieds revenue.” From a newspaper perspective, sure, that's kind of what happened. But from the Craigslist perspective, and from the user perspective, it was just like, “Well, classifieds are expensive. And you can't really get them online that well, so we're gonna make this a little bit easier for people to use.”
Following that line, if I'm a person searching for some information about the new pope, and I Google it, and I don't have to click through — casting aside some of the concerns about hallucinations and bad information for a moment — that's a good, convenient thing.
So, for news organizations, it seems like we should consider reframing our mission, or our tactics, I guess, on how to proceed in this world. How do we? How do we provide value and make money when people have to click through to our websites for these bits of information?
Mahadevan: That's a good question from an audience standpoint. Obviously, we're doing a piss-poor job of serving them if they're much happier with an automated answer from Google than they are clicking through and reading what we have to say. Obviously, there's friction — you have to click.
LaForme: I'm thinking, too, of all the news organizations 10 years ago that made a ton of money with headlines like, “What time is the Super Bowl?” And I would argue that that was a cheap way for us to earn traffic. So, I don't know, maybe this is a chance to rethink things. I'm being incredibly optimistic.
Mahadevan: I think the only way forward is to rethink our relationship to the audience. Because the way AI is changing the information ecosystem, if we don't have a good relationship with our audience and don't serve them in a way that is different from what they can be served by bite-sized, possibly hallucinated answers on Google or ChatGPT or Claude, then we deserve to die out.
Elite Truong, who was previously VP of product at the American Press Institute, had a really great session about how to experiment with AI with purpose. That's me paraphrasing it. But AI was actually not a part of the conversation. AI did not come up until, like the last 10 minutes of the session, the entire session was about the audience. How do you look at your audience as if you're fulfilling their Maslow's hierarchy of needs? Are you doing a good job of fulfilling each of those pieces of the pyramid?
LaForme: You're taking me back to freshman year.
Mahadevan: Yeah, like Psychology 101. At the bottom is: I want to have a place to sleep. Are we serving our audiences that way? Fulfilling their basic news needs? And then the very top is your sense of purpose. And are you doing a good job at fulfilling each of those pieces of the pyramid in between?
Her session was fantastic because it wasn't “Here's how to do SEO headlines,” “Here's how to create automated FOIA generators,” which are great, but it's time to rethink how we're serving the audiences in the age of AI and how to stand out from the AI slop. It seems like we actually have a really good opportunity. Because people are generally repulsed by AI.
Research on AI disclosures is finding that when people see an AI disclosure, they don't want to read on because they do not like AI. So journalists have this interesting opportunity to reconnect with audiences and show why human journalists can provide them value.
Lost in all this is a big question: What is our business? What's our business model? Why do we like search traffic? Search is good because it brings people to our site. Why do we need people to come to our site? So we can sell advertising? But what is the business? It's a complicated question. What actually is AI upending? This business model that we've been holding on to by the skin of our teeth for the past however many years. Maybe it is time that dies out. It forces us to think of new ways to make money.
LaForme: That's a delicious question and potentially a real silver lining. If you look at how things have gone since the decline of newspapers and broadcast news as these monopolies on information sources, news organizations have been following others in terms of their business model. We've not been leading. We haven't been leading for 30 years. And, yeah, maybe it's time to sort that out on our own. The status quo is in the ground. It’s cold.
Mahadevan: My dream is something like this: There's a middle aged NIMBY who lives here in St. Pete, and a 20-something who just moved here and is trying to get into the dating scene. A story comes out in the Tampa Bay Times. It's a five-paragraph story about a new restaurant — a Melting Pot is opening up on Central Avenue.
That news is important to those two people in very different ways. The NIMBY wants to complain about the Melting Pot coming in and what it means for the zoning. The 20-something might just care about the menu, or what even is the Melting Pot? How can one five-paragraph story serve those two people the same information, but in different ways?
For the young person, it might be a vertical video on Instagram created with Kapwing that's really buzzy and pulls out the very specific things that matter to them. For the NIMBY, it might be a newsletter. That's like the dream of personalization from AI.
A news story is just a piece of — I know this sounds very dystopian — but the news story is a piece of data, and then how we get it out there is what matters. The problem is, where are the humans there? And that's the question that I don't quite know the answer to. The humans have to be able to report the story.
We're at this weird period where AI is upending the information ecosystem, but also equipping us with tools that can help us do more meaningful work and connect with more audiences. AI intersects with our industry in basically every way. And wasn't that kind of like how social media did it, too? And the internet, too? It's not only changing the way we do our jobs. It's also changing the way our work gets out there in the world. And that is really scary. But it's also, like you said, a really good opportunity to rethink what it means to be a journalist, and what it means to serve people with journalism.