Google could sink ChatGPT

The text-generating AI is hailed for its lifelike scripts. But Google is declaring war on the engine
Edward Coram James
Google logo plus a stop sign

ChatGPT presents the biggest challenge that Google has ever faced. It’s a potential doomsday scenario for the search giant. ChatGPT has been shown to create false or misleading content, but content that expresses itself with such confidence that on first blush it sounds accurate. It’s also quick and cheap for web publishers to generate. If the internet becomes awash with such content, and this features prominently in Google, then Google will become synonymous with fake information and news. This, of course, would ultimately destroy trust in the search engine, which would mark the beginning of its end.

It's for the above reason that AI-generated and other automatically generated content is one of the biggest breaches of Google’s spam policies, and has been for over a decade. But it is only now that that policy is likely to truly come into play.

Google says about its spam policies:

Spam policies: The behaviors and tactics that can lead to lower ranking or being completely omitted from Google Search results.”

On that list of the top 14 spammy tactics that Google says it will penalise is ‘Spammy automatically generated content”, i.e. AI created content.

Put simply, if you use ChatGPT, or any other AI content creation software, then you may well be hit with a Google penalty, which could result in your entire website being removed from Google’s database. We believe that the internet is on course for the single biggest mass penalisation of websites in Google’s history. Websites that are using ChatGPT to generate site content will be hit hard, and any business that becomes heavily reliant upon ChatGPT for its content and thus Google related traffic risks financial ruin.

The last time such a far-reaching spam-related penalty hit was in 2012, with the Google Penguin update, an algorithm update that affected a whopping 3% of Google’s listed websites that was aimed at rooting out backlink spam. We believe any Chat GPT-related algorithm update will affect many more than this. Link spam, although easy to do, requires a degree of know-how and so was usually only undertaken by low-quality SEO practitioners. Chat GPT, however, requires no specialist skillsets and so does not require a dodgy SEO consultant to implement. Anyone with a keyboard, an internet connection and a website can generate content using ChatGPT, oblivious to the dangers, and upload it to their site.

Why does Google not like automatically generated content?

There are a number of reasons why Google dislikes AI-generated content. Principal among these is that it’s programmed to be written in a way that exudes confidence, and yet it is not fact-checked. This proliferates false information that has the veneer of authority. As Google’s entire reason d'être is to provide the most relevant, accurate and quality information to the searcher in the quickest amount of time, any software that can flood Google’s database with misleading content threatens to undermine the cornerstone of Google’s business model. If Google becomes synonymous with misinformation, then Google’s reason for market dominance evaporates.

A recent analysis of ChatGPT by John Warner shows the scale of the problem. We tested this out for ourselves at Go Up and found that, when dealing with complex subject matters, ChatGPT was prone to making subtle but significant errors. This subpar information can range from outright and blatantly incorrect to subtle misconceptions.

Writing style

Although convincingly human, the writing style of ChatGPT follows similar stylistic patterns. It loses the personality and flair of real human writers. Should the internet become awash with ChatGPT written content, then the content shown by Google runs the risk of being very bland and samey, losing the richness and interest that diversity brings.

Google could sink ChatGPT and avoid the sharp end of ordinary competition laws

ChatGPT has instantly become a direct competitor to Google’s business model and so represents the biggest commercial threat to the search behemoth since the launch of Bing in 2009. Setting aside the above issues, chat GPT has a number of advantages over Google. Its most obvious area of competition is that of Google’s ‘instant answers’, a function of the search engine wherein it answers a searcher's question without them having to click on any results, and which is also prone to some of the above issues of providing false information. ChatGPT is a clearly superior product to Google’s instant answers, providing richer, more detailed and more interesting responses to a query without the user having to click on any further links in order to obtain it.

Commercially speaking, Google has every incentive to squash it.

Fortunately for the search behemoth, whereas ordinarily competition laws would prevent them from singling out a competitor’s content for penalisation, Google’s ban on automatically generated content has been in place for over a decade, providing the perfect excuse to sink ChatGPT’s content. This would destroy the viability of one of Google’s most serious competitors, under the simple guise of adhering to its long-stated spam guidelines.  

How could Google detect content written by ChatGBT to penalise it?

This is a good question. After all, the standout feature of ChatGPT is how human-like the writing is. Do the Pepsi challenge on a piece of ChatGPT content vs a piece of human-written content, and many people will choose the AI content. But, that’s humans, not algorithms.

Google has its own chat AI software, LaMDA. And it is generally believed within large parts of the AI industry that LaMDA is significantly more sophisticated and believable than Chat GPT. Google has not released LaMDA for general use yet, largely because of their concerns about accuracy (similar issues as are facing ChatGPT) as well as the problem about the threat it poses to its core search-engine-based business model and questions on how they could effectively monetise it. But, the fact that LaMDA exists at all is the point. Google knows how to build the world’s most capable chat AI. That AI is capable of identifying and recreating patterns of writing that seem fluent and undetectable to the human eye. Google could try to adapt this same AI to be able to spot the patterns created by other AI-generated content. Building such AI spotting software into its core algorithm might be a relatively easy pivot for Google. Indeed, it is likely that their core algorithm already contains more rudimentary versions of this. As Google releases several algorithm updates every day, and around 4 core algorithm updates every year, there are plenty of opportunities for the search giant to unleash such a tweak.

How would this impact a website that has used ChatGPT to create lots of content?

If Google does decide to enforce its pre-existing rules to squash a high-threat competitor and help protect the accuracy and thus integrity of its search results pages, then this means trouble for any website that has been creating content using ChatGPT or any other piece of AI-based software. In short, if you have been building content for your site using such software, you may well wake up one morning to find that content no longer ranks in Google. If you have built your business model around this, or if you rely on these pages to generate traffic for your website, then this could spell serious trouble.

The final word

ChatGPT, like LaMDA and many other AI-based chat software, is an incredible and exciting (and terrifying) piece of technology. Many site owners and marketing agencies have already utilised it to build out instant content for their websites. Content that previously took time and money to write has been written programmatically and convincingly in seconds. But, there has not been one historical example wherein Google did not seek to penalise websites that were flouting its spam guidelines. And, get your client a Google penalty, and they won’t remain a client for long!

But, web publishers and unscrupulous SEOs are a historically short-sighted bunch. Although Google had been telling site owners and SEOs not to build spammy links for years, millions ignored this advice and made link spam one of the core pillars of their marketing strategies. So, 3% of the entirety of Google’s search results were affected, which proved fatal for many businesses.

Potentially millions of businesses will jump on board with ChatGPT, hoping that they have finally found a quiver killer SEO strategy that can short-circuit the Google algorithm and offer them nearly limitless, undetectable free content and thus Google rankings and traffic. But Google will have other plans. And those that have underestimated Google’s ability and drive to protect its commercial and quality interests have always come out wanting."

Edward Coram James is CEO of Go Up, a digital marketing agency specialising in search engine marketing and PR, operating out of London and Los Angeles. In the last 12-months Go Up has won or been nominated for some of the most prestigious awards within its industry, including the UK Search Awards, the Drum Awards and the UK Culture awards. Clients past and present include: Universal Music Group, Diageo, John Lewis, Arsenal Football Club, AnyVan, Pasta Evangelists, Huel, Hackett and Manning Gottlieb.

Written by
Edward Coram James
CEO of Go Up
December 22, 2022
Written by
December 22, 2022