For as long as Google’s been around, “authority” has been a crucial element of its success. Thanks to increasingly sophisticated algorithms, Google -or any other search engine- analyzes a vast amount of content of all kinds and determines its relevance. The famous 10 search results are the final product of this curatorship, which, with more or less efficiency, has been answering our questions for over twenty years.
Companies around the world invest millions of dollars in generating authoritative content to earn a place among those 10 results. But what is going to happen to that huge industry now that so many companies are writing their content with AI tools? How is Google going to determine the authority of this content?
Until recently, producing quality content involved hours of research, writing and rewriting, years of experience on a particular topic, and sometimes design teams that could adapt that content for multiple platforms. Today, many of these processes can be carried out with AI tools in a matter of minutes, not days. But is it the same? Can all human-generated content be replaced with AI-produced content? And what consequences could this have on the quality of content on the internet?
Innocence and Experience of AI
The quality and precision of texts produced with AI can be impressive on many occasions, but does it have the same authority as a text produced by a human expert? At the moment AI tools still make a lot of mistakes and we cannot trust 100% the content they deliver, which must be monitored. But these tools are constantly evolving: every day, millions of people use Chat GTP to make all kinds of queries, thus helping to train and improve the tool.
On the other hand, the emergence of these tools that could potentially have unlimited knowledge on countless topics urges us to rethink what it means to be an expert in something. Is it about having a degree? Having read a lot on a given topic? AI can pretend to have studied molecular biology, but is that the same as having studied molecular biology? We know that AI can digest text and learn from those patterns, just like we can. But can it replicate all the processes that constitute human learning? In a way, the interaction that AI models sustain with users constitutes a kind of experience from which the models can also learn and that over time can give them more authority on certain topics.
It’s important that we don’t mistake authority with objectivity: two experts in virology might hold different positions on the long-term effects of vaccines, for example. For this reason, we believe that it makes more sense to think of authority as a qualitative value related to someone’s journey or trajectory in a branch of knowledge. It is about the perspective that they can offer based on that professional journey. Andrew Glover, in an article in Forbes magazine, considers that AI tools: “lack the ability to make inferences and comparisons, provide experiences and anecdotes of their own, or provide a truly deep and authoritative dive into a topic.” Today we are not in a position to grant authority to artificial intelligence models. They’ll have to earn it over time, like the rest of us.
Two metrics to measure authority
Below, we detail two key concepts to understand how Google will manage to determine the authority of content in the age of artificial intelligence.
E-E-A-T
The acronym E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trustworthiness. It is a set of guidelines and evaluation criteria that Google uses to measure the relevance of the content of the sites and thus be able to rank them accordingly. The first E, which stands for “Experience”, was added at the end of 2022. This extra “E” has a very clear message: Google is saying that, faced with the rise of AI-produced content, the proven experience of a given author will be more important than ever.
These guidelines are constantly changing to accommodate the new digital reality. More than 5,150 improvements to Google’s algorithms were implemented in 2021 alone: “These changes may be big releases or small tweaks, but they are all designed to make Search work better for our users,” says Google. Reading about these guidelines and criteria, it’s clear that content that’s copied, auto-generated, or “created without effort, originality, or talent” is granted a “low quality” score. This does not mean that Google is against AI-generated content. In fact, there are plenty of experiences where Google is ranking AI-created content at the top of the results page.
Since Google doesn’t “open up” its algorithm, it’s not entirely clear whether it can detect AI-generated content in all cases. According to experiments carried out by SEO experts and enthusiasts, it is quite likely that it can spot it. As Sara Taher writes for Search Engine Land, sites that after incorporating AI-generated content suffered a drop in rankings didn’t have a lot of authority to begin with. She contrasts the case of Bankrate, a site that did have some online authority and that, when incorporating AI-generated content, did so with special attention to generating “trust signals” for Google.
What would these “trust signals” be? According to Taher, “a clear editorial responsibility disclaimer on how and where the content generated by AI is used and if it is edited by humans”, “links to external authorized sources and entities”, among others. We must note that these are just assumptions about what might constitute “best practices”: we know that Google likes to play cat and mouse with SEOs and always keep them guessing.
Information Gain Score
Another important metric to keep in mind is the Information Gain Score. It is an index that establishes how much new information a source can provide to a person who has seen other sources on the same topic. In a 2020 article, Bill Slawski already predicted how important this concept was going to be in the future. As more AI-written content appears online, Google may rely on this index to determine that the content is truly useful and not simply repeating what is already on the web. In this way, articles that provide new information will have an advantage over content that repeats or paraphrases existing content.
Source: Search Engine Roundtable
As we know, AI models write content based on the information they are fed. This means that if more and more people use AI tools to write content on a particular topic, it is possible that very similar content will circulate. To prevent this from deteriorating the quality of search results, Google’s algorithm will begin to prioritize articles by authors who offer more original content with a unique perspective, anecdotes and real-life experiences, new studies and unpublished photographs, that is, content that artificial intelligence cannot produce because it is not yet part of its database.
To conclude
As a partial conclusion, we can then say that, in the age of AI-generated content, authority and experience are still important, perhaps more than ever, as evidenced by that extra “E” that was added to E.E.A.T last year. Google will surely redouble its efforts to detect content generated with AI: not only to ensure a quality search experience, but also for basic security concerns, such as in the case of pages with especially sensitive content. Google calls these pages YMYL (Your Money Your Life) because misinformation could put people’s lives or financial security at risk. That is, pages on health, finance, food, economy, etc. Any content on the internet should be read with a critical eye: even Wikipedia can be missing key data due to lack of supervision. But in some cases, this misinformation can be very dangerous, and that’s when it becomes especially necessary for platforms like Google to intervene. We don’t know to what extent search engines will manage to contain the proliferation of cheap content facilitated by the indiscriminate use of AI. That is why our conclusion is partial: only time will tell.
- What is Digital Marketing? - February 28, 2024
- Authority in the Age of AI Generated Content - May 24, 2023
- Artificial Intelligence: a few uncomfortable questions - February 13, 2023
¿Qué te pareció este artículo?
What do you think about this post?
¡Mira también esto!
