March 5, 2024

Is Google’s AI Publisher Tool a Good Idea?

Last week, I was at BIMS—which was interesting—and then I was at the JEGI event yesterday. Next week, I’ll be speaking at the MX3 Barcelona event. If you are attending, please let me know because I’d love to meet AMO readers based out of Europe (and if you’re from the US and crossing the Atlantic with me, hit me up, too).

Speaking of events… the AMO Summit is coming back. I am choosing between two dates, so if you are planning to attend, please take this one-question survey and let me know which you prefer.

Finally, the AMO Podcast is back. You’ve all been asking for it, and I am unbelievably excited to bring the show back. Keep reading for more.

While the second half of 2023 was positive for publishers — ad rates rose, inflation dropped, and consumer spending hit record levels during the holidays — it wasn’t enough to overcome the mounting challenges the industry faces.

So, how will you win in 2024?

Sovrn’s latest data insights report examines programmatic advertising and commerce data from the second half of last year to understand the larger trends and impact on publishers. Plus, we offer top tips to help optimize for 2024, including how to run lean, future-proof against a changing ecosystem, and put AI to work for you. Download it here.

Using Google’s AI to make content

The conversation about generative AI concerning media is inherently two-sided. On the one hand, publishers are terrified of the potentially massive loss in traffic as more people move toward AI chatbots for their information versus relying on the 10-blue links. And that fear is not entirely unreasonable. According to a Gartner report:

By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents, according to Gartner, Inc. 

And while the very SEO-minded publications will suffer the most, this will be difficult for every publisher. For decades, Google has been a driver of traffic to our sites. And once this traffic hit our sites, many of us figured out ways to convert them into loyal readers. Seeing that top-of-funnel drop is going to present challenges.

On the other hand, many publishers are excited about the potential for the technology to help with various operational tasks. I’d be lying if I wasn’t using it to help with things for AMO. But one area where I think there remains uncertainty is the use of it for creating content. There are going to be some publishers that lean in aggressively to publish as much content as humanly possible for short-term gain. We saw that with Sports Illustrated back in November.

I’m not a fan of that kind of content. It’s entirely useless to the reader since it’s simply repurposing work that others have created. If not carefully edited, you run the risk of hallucination content that gives out obviously false information. As we saw with Google’s Gemini over the last couple of weeks, how these tools are built can result in bad scenarios, such as its inability to generate an image of someone white.

But is Google’s experiment with publishers the same thing? According to Adweek:

As part of the agreement, the publishers are expected to use the suite of tools to produce a fixed volume of content for 12 months. In return, the news outlets receive a monthly stipend amounting to a five-figure sum annually, as well as the means to produce content relevant to their readership at no cost.

“This speculation about this tool being used to re-publish other outlets’ work is inaccurate,” a Google representative said in a statement. “The experimental tool is being responsibly designed to help small, local publishers produce high quality journalism using factual content from public data sources—like a local government’s public information office or health authority. These tools are not intended to, and cannot, replace the essential role journalists have in reporting, creating and fact-checking their articles.” 

There are a ton of public and government sources of information that take time to look through. I use the SEC’s EDGAR and the UK’s Companies House for AMO coverage. And I suspect for a local publication, there are far more public sources out there.

So, allowing a generative AI tool to write an article based on an SEC filing would be really great. For example, BuzzFeed announced late last week that it had come to terms with its primary debtor. It wasn’t worth me taking an hour or two to write up a story on that, but could I have asked an AI tool to give me a short piece, and then I edit it? That’d probably be helpful to readers.

The trick here is to invest the time in editing. According to the Adweek story:

The resulting copy is underlined in different colors to indicate its potential accuracy: yellow, with language taken almost verbatim from the source material, is the most accurate, followed by blue and then red, with text that is least based on the original report. 

This helps the editor to understand where they need to fact-check. But with all things AI, a big part of this is the sources you put into the tool. If you use questionable sources, the output will be equally questionable.

Publishers should take this experiment one step further by using its own content to help train AI. Skift did this with Ask Skift, which I wrote about in May. At the time, Skift’s CEO Rafat Ali wrote:

We have “trained” Ask Skift on all the sum totality of Skift archives over the last 11 years, including daily stories, research reports, all of our clients’ trends reports, our specialized products – Airline Weekly, Daily Lodging Report, and Skift Meetings – and all the U.S. public travel companies’ financial SEC annual and quarterly reports. As soon as a new story or report is published, it goes straight into Ask Skift.

At the AMO Summit, Ali explained that they’ll use Ask Skift to write timeline-style articles, which are then edited. For example, this timeline is all Skift reporting, but I suspect a lot came out of its chat product. And the outcome is an excellent timeline that’s easy to understand.

Why not rely on a tool to create this kind of content? It’s suitable for the user because they can trust that the inputs are good—it is your own content and public, government sources, after all. And it’s good for you because you can create more high-quality information.

Publishers should feel comfortable experimenting with this kind of use of generative AI. It frees up reporters to focus on more important stories. That’s a positive for the reader. And we should be spending more.

Arkansas Business’ Mitch Bettis on the AMO Podcast

Mitch Bettis is the owner and President of both Arkansas Business Publishing Group and 360 West Magazine in the Dallas/Fort Worth area. In this episode, we discuss how he came to own Arkansas Business, how the business has evolved from predominantly print revenue to diversified sources, the various consumer and B2B brands he owns, and why he believes time tracking is so important.

You can listen to the podcast wherever you choose or on the AMO website. AMO Pro members also get the full transcript if they prefer to read over listening.

Click here to listen.

Thanks for reading this week. If you have thoughts, hit reply or become an AMO Pro member for an invitation to the AMO Slack. Additionally, you’ll start receiving all the content we publish each week + the Friday deep dive newsletter we send. I hope you have a great week!