Google wants to close Pandora’s box, fight AI-powered search spam


In the continual cat-and-mouse game of Google Search versus search engine optimization (SEO) firms, Google seems to be losing lately. Search feels less useful with every passing day as the ChatGPT era has unleashed a tsunami of AI junk that quickly fills up search results. Google played a big part in creating all this with its invention of transformers, and now it’s finally doing something about it. A new blog post details efforts to reduce “spammy, low-quality content on Search.”

Google’s post describes a “March 2024 core update” to the ranking algorithms that it says will show fewer results that “are unhelpful, have a poor user experience or feel like they were created for search engines instead of people.” Google says this “could include sites created primarily to match very specific search queries” and people that are “producing content at scale to boost search ranking.” The company says “based on our evaluations, we expect that the combination of this update and our previous efforts will collectively reduce low-quality, unoriginal content in search results by 40.”

Google’s post is incredibly worded not to mention AI. Google says it wants to “address emerging tactics” like “using automation to generate low-quality or unoriginal content at scale.” Google also notes that “Today, scaled content creation methods are more sophisticated,” but which new “content creation methods” spammers are using is left as a mystery. Google wants to style itself as an AI-first company now. Apparently, that means never directly mentioning any of the downsides of the AI-powered internet Google played a role in creating.

Google’s response comes after there have been several high-profile search spam stories lately. One viral X post bragged about an AI-powered “SEO heist” where a website copied a competing website with AI, was able to rank more highly than the site they copied from, and, by their own admission, “stole” 3.6 million page views. A recent study from Leipzig University, Bauhaus-University Weimar, and the Center for Scalable Data Analytics and Artificial Intelligence claimed Google was losing the war with SEO firms. Another post from product review site HouseFresh detailed how Google doesn’t actually prioritize quality review articles and instead just lets big publishers spam low-quality “best of” affiliate link articles to the front pages of the search results.

Google’s vague statement that it might kill off sites “created primarily to match very specific search queries” could be interpreted as a shot at the world of affiliate link articles. Google also says it wants to stop “site reputation abuse” where “websites that have their own great content may also host low-quality content provided by third parties with the goal of capitalizing on the hosting site’s strong reputation.” (If this means I get less email spam for “Ars Technica guest posts,” that would be great.) Another change thrown into the pile is “expired domain abuse: where recently expired domains will be downranked.

It’s hard to know what Google considers “low-quality content.” Google’s policies still don’t penalize AI-generated websites, and just the other day, it was caught paying news sites to create AI-generated articles. As a user, I feel like if I wanted to see AI content, I could just go to an AI thing and generate it myself. Using Google Search in the past has always meant you were looking for articles written by humans (admittedly using varying degrees of effort), and to me, it makes sense to keep it that way. However, Google is still reluctant to ban AI altogether (again, it wants to be an AI company now).

With AI quickly becoming ubiquitous across the Internet—including at the top of Google Search if you turn it on—having the search links meet stringent quality standards seems like soon it will be their only point of differentiation. In the new AI world, if Google isn’t aggressive enough with search quality, it risks just losing users. If the first 10 blue links aren’t up to snuff, people might just ask ChatGPT instead.



Source link

About The Author

Scroll to Top