In November 2021, we shared our belief that Web3 would be powered by AI with a private business community group hosting over 13,000 members. Our post faced intense criticism and backlash. Two years later, in February 2023, both Microsoft and Google, who already leverage machine learning and AI algorithms in their search engines, are publicly introducing search-augmented generative models as a companion tool.
This recent surge in popularity of generative models and the integration announcements made by both Microsoft and Google have sparked a debate among digital marketing practitioners. Some believe that these developments, including the ability to generate AI content on a large scale, could fundamentally transform our understanding and approach to Search Engine Optimization (SEO). Our view is that, at present, the limitations imposed by vendors and current legal considerations, combined with the current capabilities of generative models, will not bring about a significant impact on Search Engine Optimization (SEO).
How do search engines rank content?
Search engine ranking algorithms determine the relevance and quality of a website or web page based on its content and other factors, such as the popularity of the website and the number of high-quality links pointing to it. These algorithms use a complex set of criteria to determine the ranking of a web page, including keyword analysis, on-page content, website structure, and user experience metrics including bounce rate, page load speed and time on page. The ultimate goal of these algorithms is to provide the most relevant and useful results for a user’s search query, ensuring that the user finds the information they are looking for quickly and easily.
AI generated content at scale
The advancement of AI technology has made it possible to generate content at scale. AI writing assistant tools based on generative models such as GPT, use NLP and machine learning algorithms to allow the creation of original content without human intervention. These models have been trained on vast amounts of text data and can generate articles, reports, and other types of content based on specific prompts or criteria.
Over the past two years, some writers have experimented with generative models, such as GPT3 and more recently GPT3.5, as well as other AI writing tools, to significantly increase their content production through mass production or automation. Some of these writers boasted about their successes, showcasing traffic spikes in their analytics on community groups. However, many of them disappeared shortly after and reported no improvement in their overall site search engine rankings.
Out-of-the-box generative models lack accuracy
One of the the main reasons why a lot of these writers did not experience the type of drastic change in rankings they were looking for, is that generative models lack the accuracy and precision needed for high-quality content when used out of the box without extra human editing. These models are not accurate if used within a closed-book system, where they are queried directly and respond using their internal data only. In these scenarios, the model may generate answers that are not rooted in reality, and the information it provides may be incorrect outdated or misleading. To address this limitation, many individuals have opted for fine-tuning models with more domain-specific data or incorporating the generative model into a retriever-augmented pipeline. However, these solutions often incur high training or deployment costs that are not financially feasible for most people.
Vendor and legal limitations
It is also worth noting that both Microsoft and Google are offering search-augmented generative models exclusively through their search engine pages, not through third-party service providers or APIs. The API versions available are not search-augmented, which speaks volumes. While it is possible to create an alternative search-augmented data pipeline using multiple pipeline elements, however usage rate limits on Open AI API calls remain capped and will not change in the near future. The recent release of a tool by OpenAI that can detect AI-generated content can be seen as a clear indication of the vendor’s own views on the potential widespread use of AI-generated content. These limitations come as a direct result of both commercial considerations, but also other challenges associated with search-augmented generative models that the vendors might be concerned about, such as the sources used for generation (human-created or machine-created), potential copyright infringement, platform liability under Section 230 of the Communications Decency Act, and quality degradation of machine generated content when used in a chain.
What about online shoppers?
Despite these advancements in AI-powered technologies, shoppers’ behavior is unlikely to change in the near future. Consumers will still visit e-commerce sites, browse products, and compare prices, just as they have done in the past. The advent of aggregators has had a significant impact on e-commerce, but it has not completely transformed the way people shop. This means that e-commerce sites will continue to strive for better search engine rankings and work on their SEO strategies. In a highly competitive online marketplace, companies must constantly optimize their search engine rankings to attract more customers and increase visibility. As such, search engine optimization will continue to be an important aspect of e-commerce, regardless of the advancements in AI technologies.
The future ahead
While the SEO landscape is constantly evolving, it is unlikely to change drastically in the near future and in response to search-augmented generative models due to technical limitations, ethical and legal concerns, and the behavior of consumers. The importance of content quality in search engines cannot be overemphasized, and it is essential for websites to keep focusing on providing high-quality, accurate, and original content to improve their ranking and visibility.
ABOUT THE AUTHOR(s)
Roland Tannous is managing partner and lead strategy and digital transformation consultant at GravityThink.