Google has abandoned the notion of "human-written content" and now says it's looking for quality content - no matter how it's generated.
Google has long preached the idea of "content written by people for people". But in a recent update, the giant is quietly rewriting its own rules, acknowledging the rise of artificial intelligence.
In the latest iteration of Google's "Useful Content Update," the search engine's phrase "written by humans" has been replaced with the statement that Google constantly monitors "content created for humans" to rank sites on its search engine.
The new wording shows that Google recognises artificial intelligence as a tool that is heavily relied upon for content creation. But instead of simply focusing on differentiating AI-created content from human-created content, the leading search engine wants to highlight valuable content that is useful to users. Regardless of whether it was created by humans or machines.
Meanwhile, Google is investing in AI across all its products, including an AI-powered news generation service alongside its own chatbot with Bard and new experimental search features. Updating the guidelines is also in line with the company's strategic goals.
The search leader continues to strive to reward original, useful and human content that provides value to users.
To do SEO or not to do SEO?
The implications are clear: repetitive or low-quality AI content can still hurt SEO, even as technology advances.
Copywriters and editors still need to play an active role in the content creation process. Lack of human involvement is risky because AI models tend to hallucinate. Some of the mistakes may be funny or offensive, but some of them can cost millions and even endanger people's lives.
SEO, or search engine optimization, refers to strategies aimed at improving a website's ranking on search engines like Google. Higher rankings lead to more visibility and traffic. SEO experts have long tried to "beat" search algorithms by optimizing content to match Google's algorithm.
Google seems to penalize the use of artificial intelligence for simply summarizing or paraphrasing content and has its own ways of detecting AI-generated content.
How Google detects AI-generated content
"This classification process is fully automated using a machine learning model." says Google, meaning it uses AI to distinguish good from bad content.
Part of the challenge, however, is that AI-generated content detection often relies on inaccurate tools. OpenAI itself recently removed its own AI classifier, acknowledging its inaccuracy.
AI detection is difficult because models are actually trained to "look" like humans, so the confrontation between content creators and content distinguishers will never end because AI models only become more powerful and more accurate over time.
Furthermore, training AI with AI-generated content can lead to model collapse.
Google says it's not trying to replicate AI-generated data, but rather identify it and reward human-written content accordingly. This approach is more like training a specialized AI discriminator, where one AI model tries to create something that looks natural and another model tries to discern whether the creation is natural or artificial. This process is already used in generative adversarial networks (GANs).
Standards will continue to evolve as AI proliferates. For now, Google seems to be focusing on the quality of content rather than separating human input from that created using machines.