It’s been a tumultuous couple of weeks for the search industry at large. Between the Yandex leak, Bing’s native integration of ChatGPT into search and Google’s rushed unveiling of the most talked-about Bard since Shakespeare, there’s been a lot to take stock of.
Indeed, if the average LinkedIn feed is a microcosm of the SEO industry as a whole, it’s clear that recent AI developments are being seen as the biggest disruptors to search since… well, ever.
But how does Google’s release of Bard - and Bing’s ChatGPT-powered rebirth - reconcile with the commonly held, and often perpetuated, idea that AI content is objectively bad?
You’d probably have to have been living off the grid for the last few months if you’ve yet to come across some of the online chatter surrounding the launch of OpenAI’s ChatGPT chatbot.
With a phenomenal reception due to its human-like responses and powerful understanding of natural language, ChatGPT has already seen hundreds, if not thousands, of commercially valuable use cases since its prototype was released back in November.
Naturally, this has led many within the SEO industry to consider - and agonise about - whether the era of rankable AI-generated content is truly upon us. While the prospect of automated content of any form outperforming human copy has long been met with scorn, there is definitely a feeling in the air that things might be different this time; a feeling cemented by how search engines themselves have reacted to ChatGPT capturing the imagination of the world.
Initially leaked at the start of January and confirmed at the end of the month, early OpenAI investor Microsoft moved their partnership to the next level with a $10 billion investment in ChatGPT.
While many initial predictions have come to fruition already, including plans to integrate ChatGPT into some of Microsoft’s flagship productivity tools such as Teams and Office, many expected it’d be some time before the tech giant looked to incorporate some form of AI functionality into its own search engine, Bing.
Microsoft’s announcement late on Tuesday (February 7) that it was set to immediately roll out a new, ChatGPT-powered version, of the Bing search engine and Edge browser into open preview, therefore, came as a widespread surprise.
In a press release, Microsoft’s Yusuf Mehdi wrote of Bing’s ability to provide “complete answers” to the more complex queries users have, using the example of “how to substitute eggs for another ingredient in a cake you are baking” being answered directly in the SERP to illustrate the benefits of ChatGPT-powered search.
Outside of this, we read promises of “better search”, “a new chat experience” and the ability to add a “creative spark” to users’ search queries.
It certainly seems that Microsoft, whose distant-second search tool Bing has been quietly gaining market share of search over the past few years, have struck the first meaningful blow in the AI search arms race.
As early as December, it was reported that Google was taking ChatGPT very seriously indeed. With one eye no doubt on Microsoft and OpenAI’s strategic partnership, CEO Sundar Pichai allegedly redeployed various teams to focus on Google’s own artificial intelligence programmes at a rapid pace.
Fast-forward to last week, and CNBC reported that Google was mass stress-testing its own ChatGPT challenger, then known as Apprentice Bard, under “code red” conditions with a view to an upcoming product launch.
With Microsoft’s Bing announcement imminent, Google moved quickly to upend it with public confirmation of Bard’s existence and a sneak peek at how it will integrate directly with search in the near future. As it turns out, Google moved too quickly.
Despite acknowledging the “reputational risk” associated with inaccurate AI content, Google was found to have highlighted a misleading answer from Bard on a NASA telescope in its own promotional materials for the engine. An inaccuracy that as of the time of writing, has wiped 7% - or $100 billion - off the value of Google parent Alphabet’s shares.
That said, there is little doubt in my mind the search giant will bounce back soon enough, but it does speak volumes about the volatility and uncertainty surrounding AI content in 2023.
Those in the SEO content industry may have been more surprised than most to see Google move so quickly to challenge ChatGPT. After all, while its LaMDA and MUM (previously BERT) AI technologies have long been touted as a natural evolution in search, Google has historically been categorical in advising web publishers to avoid automated and AI-driven content of any kind if they want to perform well in search:
This was often echoed by prominent Googler staffers such as John Mueller, who advised site owners that AI tools were useful for some aspects of content strategy - but not production.
However, eagle-eyed SEOs spotted a change in tact as early as April 2022, when Google updated its guidelines to only dissuade webmasters from publishing AI content that is explicitly intended to manipulate search rankings.
Since that time, the company has doubled down on producing “content for humans” rather than search engines, while deliberately avoiding instigating a complete no-go zone for AI-generated copy.
In an explainer posted Wednesday (February 8), no doubt to coincide with its Bard announcement, Google provided more explicit guidance on AI-generated content. Notably, it’s keen to promote the idea that content is content - regardless of how it is produced.
Essentially, as long as a website is able to satisfy criteria such as E-E-A-T and Google’s Helpful Content system, it doesn’t necessarily matter how that content is produced. The blog post also provided one of the largest softening-of-tones we’ve seen from the company in the context of AI-generated content for search:
“AI has the ability to power new levels of expression and creativity, and to serve as a critical tool to help people create great content for the web.”
Reading on, Google reiterates that “appropriate” AI content itself is not against its guidelines; that is, if automation is used to create content it considers “useful”, that’s fair game. The issue comes, as was explained last year, when sites use automation to create spammy content that exists to “game” the system rather than provide real value to users.
The company also goes on to reassure readers that its systems to detect inaccurate and misleading information are equally as valid for AI-generated content as content created by humans. But as we’ve seen with Bard even before its official release, Google is not infallible.
Reading all of the above, you might be thinking that the time is nigh to ditch human copywriting entirely in favour of embracing AI. Bard issues notwithstanding, however, there is enough evidence to suggest that AI content isn’t - at least yet - a feasible replacement for human-produced copy.
A number of tools, such as AI Text Classifier from OpenAI themselves, have proven to demonstrate a high ‘detection’ rate for AI-generated text. Of course, as we now know, the existence of AI text in itself isn’t officially an issue in terms of search rankings.
However, the proven fallibility of AI text and extent to which it can be currently recognised as distinct to expert-written copy suggests it would struggle to satisfy Google’s stringent E-E-A-T guidelines at this point in time.
Elsewhere, empirical evidence has shown that websites that have had an overreliance on AI-generated content have struggled long-term in the SERPs, but also - even more critically - with the goodwill of their readers.
In a recent experiment by prominent marketer Neil Patel, it was demonstrated that websites leaning on purely AI-generated content saw an average 17% decline in rankings, with a typical eight dropped keyword positions.
While my view is clear insofar as that it’s not going to make copywriters and SEOs redundant any time soon, AI in search is most definitely here to stay. With Google and Microsoft clearly “all-in” on their respective solutions - no matter their drawbacks - that means that we as a search industry should embrace the technology wherever it can enhance and make efficiencies to the content production process.
Rather than seeking to replace the currently irreplaceable in the human element of copywriting, consider how to leverage the power of AI to inform the earlier stages of content creation - from assisting in keyword research and clustering, to partial metadata automation, topical research and structural support for your articles.
At ICS-digital, we’re naturally proud of our expert international content capabilities, and believe there is no substitute for the deep understanding our copywriters have of their target audience. But we’re also not afraid of embracing AI to innovate and improve on our overall workflows and processes where it makes sense to do so.
Get in touch to find out how we can develop and deliver your SEO content strategy, fit for the era of AI-powered search.