Should brands funding journalism be happy running ads in AI-generated articles?
Cnet and BuzzFeed are now investing in AI-generated content. Will the advertisers intent on supporting quality journalism be happy to pay a premium to appear next to these stories?
What impact will AI-generated content have on digital ad pricing?
The news that BuzzFeed was set to use ChatGPT to create a portion of its articles was positively received by the markets. Off the back of the announcement, the digital brand’s stock price rose 120% as investors foresaw an opportunity for the AI tool to boost BuzzFeed’s content capabilities – and therefore its advertising opportunities.
The idea is attractive to markets: BuzzFeed is cutting the costs of creation while dramatically increasing the number of ad spots available to buyers. But for marketers, there are several questions that need answering - not least of which is ‘how many other digital publishers will follow suit?’
One potential issue that has already reared its head is that of brand safety. In early January, financial news and advice site Cnet announced that it had been quietly using AI to create articles for some time and that it intended to continue ramping up that practice. Less than a fortnight later the site was forced to hit pause on its proprietary AI tool, however, as users noticed that basic errors were slipping through and because the site had not adequately disclosed which articles were AI-generated.
That single example raises the issues of brand safety and user perception surrounding AI-generated articles. With many publishers trying to frame their sites as being premium spaces with commensurately greater CPMs, the use of AI-generated articles throws that ‘premium’ distinction into question.
Will Davis, senior SEO specialist at Vertical Leap, said publishers could have to rethink the pricing of ads on AI-generated content.
“Where performance-based costs are concerned, a typical advertiser is interested in users moving onto their site or becoming aware of their brand or product. I believe that we will see an increase in the rate at which content is generated as AI takes hold, but if the ad is achieving the same outcome, then there is no reason to have the ad cost directly reflect the labor that went into the page,” he explained.
“Of course, if we see ads performing less effectively on sites that implement this approach, then that is something that should be reflected in the cost.”
Crowd copywriter and AI-generation specialist Dan Tull agreed that the pricing of display ads served opposite AI-generated articles may differ based on factors such as the perceived quality and audience engagement with the content. “Ultimately, I think it’ll end up being another consideration for marketers, just like the interplay of paid and organic media, AI could have its own space,” he said.
Brand safety and SEO
Additionally, the issue of brand safety surrounding AI-generated content has been highlighted multiple times over the past week. While ChatGPT has restrictions on which prompts it can and will respond to – explicitly to prevent brand safety issues surrounding the incitement of violence and illegal activity, racial slurs, etc. – those restrictions have already been circumvented.
On Monday, another generative AI got its user into hot water when the AI-created Seinfeld series ‘Nothing, Forever’ was given a 14-day Twitter ban for creating transphobic content. The implication is that human oversight is still required.
Henry Carless, PPC and data science specialist at Vertical Leap, explains: “If we've learned anything from Microsoft’s ‘Tay’ Twitter bot, or the Seinfeld AI on Twitch – both of which became suddenly and extremely racist – we know that AI can go off the rails very easily.
“That said, brand safety is difficult enough as it is, and in the grand scheme of things I can’t see it having a major impact in the short term. I’d rather show ads on an AI-generated news article than a flashlight app or a YouTube channel with no subscribers and a 20% Adsense click-through rate.”
Perhaps the most important consideration for any publisher looking to use AI-generated content is the impact it will have on its search strategy. Google has been deprioritizing AI-generated content on the grounds of ‘quality’, already causing an arms race between AI platforms and the search giants.
Andy Cove, digital marketing director at M3.agency, said: “We do see some red flags and issues starting to be raised. There are already tools that predict the likelihood that content has been generated by AI so we’re likely to see content flagged and penalized by search engines and other big tech companies. This highlights a risk for BuzzFeed and its advertisers.”
That is a risk that’s exacerbated by a potential rise in bad actors using AI tools like ChatGPT. In an experiment, Mahadevan created an entire news website and its staff “out of thin air”, illustrating the need for the industry to get to grips with what AI-generated articles will do to pricing. As a result, media buyers and SEO specialists believe that human originality will be the proof point for sites claiming to be ‘premium.
James Ellis is strategy director at JWE Media. He explained: “Time will tell [that] brands will have to ensure that their content is still as engaging and has a human element/personality. If everything is AI generated the content and tone of voice will be very similar across various digital brands and their content won't have any personality.”
With Microsoft and Google both set to implement AI tools more deeply into its own search tools, it is evident that 2023 will be the year in which AI really comes into play for digital advertising - both in terms of creation and monetization of the content against which it is served.