Here’s How Google Search Created the Problem of AI-Generated Content

  • Google appears to be down ranking pages created by AI that it deems low quality.
  • Experts predict the web will be flooded with AI-generated content as the technology improves.
  • Google’s internal technology may evetually not be able to detect whether or not a page was written by an AI.

A few weeks ago, online marketer Neil Patel began noticing that some of the websites he tracks were seeing major declines in traffic – some by as much as 70%.

The websites were test pages he had created and populated with articles generated by artificial intelligence. And the declines came in the wake of an update to Google’s search algorithm from October designed specifically to root out spam. It was a follow-up to an update from August where Google said it was trying to down rank content it deemed “unhelpful.”

Patel, who blogged about the experiment, reached the conclusion that Google had finally come for the robots and their AI-generated content.

It’s a significant development for the web and roiling controversy in the world of web traffic, where scores of Search Engine Optimizers (or SEOs) watch and analyze all the small moves that Google makes to its search algorithm so they can advise clients on how to appear near the top of search results. But if Google is indeed declaring a full-scale war against AI-generated content, it may be swimming against the currents of where content creation is headed — and testing the limits of the company’s own AI detection abilities.

Experts say the web will soon be awash in content that is AI-assisted—if not created by the technology entirely. Companies with websites that depend on Google traffic have increasingly relied on AI to write content. It’s particularly common for e-commerce websites, which often have blurbs about hundreds of products. Using the tech to do write ups on the different items so they can be indexed in a Google search cuts down on the costs of farming the work out to copy writers.

The trend has major implications for the future of the web, according to tech executives and search engine specialists. The quality of AI-generated content, including text created with startup OpenAI’s technology GPT-3, is already often good enough to fool human readers. If Google did nothing, the majority of pages that people visit through Google search could one day feature content written only by AIs.

“AI has the potential to change search as we know it,” said Sridhar Ramaswamy, a former senior vice president on Google’s ads team and co-founder of search startup Neeva. The signals that Google relied on for quality like the domain and links to it “are becoming less and less relevant,” he added.

For now, Google representatives haven’t taken a strong stand on AI generated content. In a statement to Insider, Google’s search liaison Danny Sullivan said the company’s recent updates to search are focused more on weeding out pages that were created solely to game the algorithm rather than penalizing pages that were made with AIs.

“Our systems focus on the usefulness of content, rather than how it’s produced. This allows us to consistently provide high quality results, while reducing all forms of unhelpful content in Search, whether it’s created by humans or through automated processes,” Sullivan said.

But as the technology improves and AI generated content becomes more commonplace, Google will be increasingly challenged to tackle the topic more directly. Google says in its search guidelines that it looks for areas like expertise, authority, and trustworthiness when it comes to ranking pages. For example, when a user searches “how to improve a credit score,” Google will favor a site that’s been publishing content about financial advice for years over a new site that has a similar answer to the question.

Even if writers use AI in the Google-sanctioned way by creating helpful content with a human’s touch, it could create a giant rush to make articles as similar as possible. The AI ​​could fine tune writing to make it search engine optimized, leading to too many articles with similar content and quality, said Mike King, an online marketer and blogger.

“If everyone can create a perfect version of an article, how do you as Google determine which one should rank?” King said.

Even if Google wants to down rank AI-generated content, the growing sophistication of the technology could make that hard to do. The next model of generative AI will be higher quality and trickier for an algorithm to discern from human-generated text.

Google still has the advantage for the time being. In the AI ​​versus AI arms race, the internal technology at the search giant is far more sophisticated than what’s available off the shelf, and that will keep it ahead in detecting AI-generated text, said Jeff Coyle, chief strategy officer at AI content company MarketMuse.

Of course, if Google does view AI as a major threat to search quality, it only has itself to blame. As one of the largest backers of AI research and development, the company has done more for the technology than almost anyone else in the industry.

“It’s ironic that they have to even think about this because they effectively created the problem,” King said.

Do you work at YouTube or Google? Got a tip? Contact reporter Tom Dotan via email at tdotan@insider.com or Twitter DM at @cityofthetown. Check out Insider’s source guide for more tips on how to share information securely.

Leave a Reply

Your email address will not be published. Required fields are marked *