How AI-generated content challenges SEO and web operators

The past few years have been tumultuous for search engine providers, with Google, Bing, Yahoo!, Yandex, DuckDuckGo, and the rest having to adapt to the impact of AI, machine learning, and large language models.

My colleague Richard MacManus has already written about the future of websites in the AI ​​era and how search engines are turning into generative AI chatbots. Factor in declining referrals and diminishing social media traffic to websites, and the list of potential headaches for SEO practitioners seems to grow.

That’s not all: the advent of AI has made it remarkably easy to produce content of all varieties, from the written word to still images and video. With a veritable flood of AI-generated content being pumped out onto the Internet at an ever-increasing rate, how can search providers help readers distinguish between the gray sludge of AI-produced drivel and the valuable nuggets of content on the Internet, ideally produced – or at least curated – by real people? As someone who has spent most of his adult life creating, editing and strategizing the content of the man-made variety, I admittedly have a vested interest in the subject.

AI Content Spam Conundrum

While it is undeniable that AI can be a valuable aid to content producers – helping to generate ideas, transcribing content from one format to another, quickly generating copy that can be edited and refined later, among many other useful examples – the use of ​​AI to generate low-quality content is already having an impact, especially when used in ethically questionable ways. Here are a few recent examples:

Former tech journalist (and current senior developer advocate at Github) Christina Warren learned that the current owners of TUAW bought the domain and then “…populated it with AI-generated slop and then reused my name from a job I had when I was 21 years old to try to pull some SEO scam that won’t even work in 2024 because Google changed its algo.”

Images, illustrations, and video content are not immune to the AI-generated slop attack.

Then there’s the story of an SEO agency that boasted of carrying out an “…SEO heist that stole 3.6M total traffic from a competitor”, which involved exporting the competitor’s sitemap, converting the… their list of article title URLs…” and then create “…1,800 articles from those titles at scale using AI.”

Alex Hern and Dan Milmo at The Guardian have described efforts by some in the tech community to designate this new category of AI-generated content as Slop rather than Spam, while Reece Rogers at Wired has discussed how AI spam ranked above the original reporting in news results.

Images, illustrations and video content are not immune to the AI-generated slop attack, as Shannon Bond at NPR reported on AI-generated images beginning to flood social media feeds. Bond wrote that “…in addition to finding AI spam on Facebook annoying, many people NPR spoke to say they are concerned about the larger efforts of artificial images popping up everywhere.” Note the concern expressed here about AI generated content. I’ll elaborate on that concern and what it might mean for all of us in a bit.

The impact of AI on search and content

To get an additional perspective on how AI is impacting search, SEO, and content development, I reached out to Duane Forrester, senior vice president of search at INDEXR.ai—and the former head of the Bing Webmaster Tools program—to get his take on the positives and negatives that AI brings to the table for search, SEO and content efforts.

“It’s like having access to the Internet’s library of technical advice and content creation advice within one system.”
– Duane Forrester, INDEXR.ai

Forrester praised the increased efficiency that AI tools can provide, from faster workflows to increased throughput. “[I’m] thinking about content here, especially when approached from the idea generation side, but also from the pure writing side… [AI also provides] increased brand voice customization, [as it’s] easier to tell a system to write a certain way and have all baseline drafts adjusted in voice… [and also] makes editing basics like grammar and syntax easier when creating content.”

Another bonus is the power that AI brings to optimization work. “An AI-based system can… look at your pages, your code, your content and offer recommendations based on the universe of known best practices,” Forrester said. “It’s like having access to the Internet’s library of technical advice and content creation advice within one system.”

On the negative side, Forrester notes that some companies think they can replace writers, but notes that “…those who have often suffered.” In addition to the impact on content producers, Forrester notes that “… the flood of low-quality content not only affects search results, but when consumers engage with these poorer results, it trains both the consumer and the search platforms to feed more of the low-quality content into the results as people engage with it.”

Taking the impact of low-quality content a step further, the ease of generating AI-produced content can lull businesses into a false sense of security. “Companies mistakenly assume [that] whatever an AI generative system produces is high enough quality to skip review and just publish,” Forrester said. “The lack of oversight can pose a serious problem if you publish low-quality, inaccurate or dangerous content.”

Fixing the problem: Registration of AI content

So what are Google, Bing, and other search engines doing about the problem of easily generated AI fluff clogging up our search results? Both Google and Bing have released updates or made improvements to their algorithms to address issues, and will no doubt continue to improve how their search engines handle AI-generated content.

Search engines “remain diligent and aware of bad actors trying to link spam to domains, and they thwart these instances.”
– Forrester

“From personal experience at Bing, I know that employees show up every single day with the sole focus of their work to spot and stop bad actors,” Forrester said. “Google has similar teams. All focused on quality results and security. I can say with 100% confidence that both of these big engines are taking this seriously and working hard to solve this problem. Can they now do more? Absolutely. There are, I’m sure, countless ways spammers can still exploit things that are known and should be dealt with. Yet these problems remain, still unsolved, and still allow the problems to continue. The engines will tell you that ‘negative SEO’ does not is one thing. That doesn’t happen. They stay diligent and aware of bad actors trying to link spam to domains, and they reliably thwart those instances.”

The AI-Generated Content Trust Gap

Remember a few paragraphs ago when I said I wanted to revisit the concerns consumers have about AI-generated content? While some companies are using AI as a valuable tool to help existing employees become more efficient, others have unfortunately downsized with the idea that AI can take over. While the efficiency gains AI tools bring as support tools are undeniable, it’s becoming clear that AI-generated content also has a credibility problem.

62% are “less likely to engage with and trust content if they know it was created by an AI application.”
– Hootsuite Report, 2024

In addition to the cautious consumers who fear AI-generated content that I referenced earlier, a bit of human-driven research reveals that they’re not alone: ​​According to HootSuite’s Social Media 2024 Trends Report, 62% of survey respondents indicated that they were “less likely to engage with and trust content if they know it was created by an AI application.” Throw in the news of AI-generated articles by fake writers at Sports Illustrated and USA Today, and you can see why many consumers are wary of what AI means for the content they consume.

There are also ongoing concerns about AI “hallucinations” – including a widely reported case of Google’s AI Overviews feature suggesting that people “…make pizza with a quarter cup of non-toxic glue as an ingredient to prevent cheese from sliding off the pizza after cooking.” Google has since fixed that specific problem, but it’s sound advice to treat AI as a useful tool that still needs hands-on supervision by experts human employees, rather than a fire-and-forget panacea that narrow-minded CEOs believe can replace their search and content marketing departments wholesale.

The Future of SEO, AI and Content Spam

Google and other search providers are clearly making a concerted effort to improve how their search algorithms find and reward quality content of the AI ​​or human variety, with Google’s latest Broad Core update from August 2024 continuing Google’s efforts to “…improve the quality of ​​our search results by showing more content that people find truly useful and less content that feels like it was made to do well on Search.”

“If you know your customer inside out, you’ll be in the best place to deliver exactly what they need, making you invaluable to the search engines.”
– Forrester

Google has repeatedly stated that it intends to reward quality content, ie written for humans over search engines – regardless whether the content is created by a human or AI– So time will tell how successful Google, Bing and other search providers will be in separating mass-produced AI content spam from quality material written (or at least guided by) human authors.

When it comes to other advice for SEO and content practitioners, Forrester suggests staying focused on their customers’ needs first. “It may sound trite today, but who do you think the big search companies are focusing on? That’s right… your customer. If you know your customer inside out, you will be in the best place to deliver exactly what they need, making you invaluable to the search engines.”

In addition, Forrester recommends keeping a close eye on all the ongoing changes to search and AI tools as they are updated and improved over time. “We will see more and more frequent changes going forward. Missing them and not responding can cost a company, so stay tuned and understand how the changes affect consumers, then adjust as needed.”