AI-powered search tools like SearchGPT have sparked both excitement and concern among SEOs. Recently, industry prognosticators have expressed fears about AI-driven tools undermining traditional traffic sources. They worry that widespread adoption of these tools will divert so much traffic away from previously profitable websites that the decline in visibility will “unalive” those sites completely.
It’s normal to feel nervous about new technologies, especially when they seem to change not only the core of how you understand search works, but also change searcher behavior in general. Thankfully, anxiety fades when facts are clear. Let’s break down why AI-powered search is not built on theft and how it fits into the broader SEO ecosystem.
Understanding how AI-powered search works
Before addressing the claims of content theft, we must first understand how tools like SearchGPT function. At its core, these AI tools are large language models (LLM) trained on vast amounts of publicly available text data. This training process involves learning patterns in language to generate human-like text responses. Unlike traditional data analysis or fact-learning, the “training” focuses on understanding and predicting language rather than memorizing specific facts. This, then, begs the question: “Does it really, truly answer questions, or does it just create accurate-sounding guesses?”
Data gathering and synthesis
When a user submits a query, SearchGPT (specifically, but this is probably true of similar tools) processes the input by analyzing and interpreting the request using its trained language patterns. This means the way the search for facts is constructed is more accurate than relying solely on the terms inputted by the user. Then, instead of merely searching for and retrieving existing content, it looks at all the content from multiple top sources. It synthesizes that information (meaning it reads and evaluates all the retrieved content) to construct a coherent and comprehensive response.
The “synthesizing” process involves:
identifying relevant data points
understanding the context of the query
evaluating the expertise and likelihood of accuracy
and integrating information in a way that aligns with the user’s intent
Generating original content
SearchGPT doesn’t copy and paste text from websites. Instead, it generates new content based on the patterns it has learned during training. This process is similar to how human writers use their knowledge and experience to create original articles. By leveraging sophisticated algorithms, SearchGPT (and other tools) ensures that the generated text is both unique and informative, providing valuable answers without replicating existing content verbatim.
Ensuring accuracy and relevance
To maintain high standards of accuracy and relevance, AI-powered search has processes in the background to evaluate the reliability of the information it synthesizes. It prioritizes data from authoritative sources, cross-references information to minimize errors, and continually adapts to new information to provide up-to-date answers. This dynamic capability ensures that users receive responses that are not only accurate but also reflect the latest knowledge and trends.
The dynamic nature of AI responses
This is where the general understanding of how AI creates search responses and content goes astray. When tools like ChatGPT or AI-powered search tools write new content or provide answers, they are not relying solely on data points and facts that they have been *previously* trained on; the data set is not “old.” The AI’s ability to search for and adopt new information allows it to refine its responses over time, ensuring that the answers remain relevant, valuable, and accurate. This continuous learning process means that AI-powered tools can adjust responses to better meet users’ needs as user behavior and information evolve, providing a more personalized and effective search experience.
To sum up, tools like SearchGPT function by gathering and synthesizing information from a wide array of sources (that they find on the web) to generate original, accurate, and relevant responses to user queries. This process ensures that while the AI provides quick and concise answers, it does so by leveraging a deep understanding of language and context rather than stealing or copying content.
The role of attribution and source linking
The biggest concern among SEOs is the traffic loss if AI-powered search provides direct answers without driving clicks to the source website. This fear is completely understandable but overlooks an important aspect: the role of attribution.
Many AI-powered search engines, including those integrating models like SearchGPT, prioritize providing users with accurate, high-quality information. In doing so, they include links back to the original sources. This attribution ensures that websites receive credit and traffic for their content. Rather than stealing clicks, AI serves as a conduit, directing engaged users to the source of the information for more in-depth exploration.
However, this also means that the core concern, that some sites might lose traffic because they are not chosen as the cited source for the information, is a reasonable concern. The way to combat this loss isn’t to fight against adopting new technology (because that is probably futile at this point). The best way to fight is to ensure the cited source is YOUR source. This will become the new focus of SEO.
AI as a complement, not a competitor
The notion that AI-powered search will “kill” websites is rooted in a misunderstanding of how these tools are intended to function. AI doesn’t replace the need for high-quality, authoritative content; it amplifies it. Search engines and AI tools rely on the experience, expertise, authoritativeness, and trustworthiness (EEAT) of websites to deliver relevant and credible information to users.
Websites that invest in EEAT will continue to thrive, as AI tools will naturally prioritize their content in response to user queries. In this sense, AI becomes a partner in the SEO journey, helping to surface the best content and ensure it reaches the right audience.
Websites that are “killed” by AI-powered search won’t be innocent victims of a new technology run amok; rather, they’re more likely to have been removed from the knowledge pool by digital Darwinism — “survival of the fittest“ and all that.
The future of SEO in an AI-driven world
As with any technological advancement, AI-powered search tools will require SEOs to adapt and evolve their strategies. However, this evolution doesn’t mean the end of traditional SEO practices. Instead, it highlights the importance of optimizing for both AI and human users — which is not all that different from the old guidance to “optimize for both robots and human users.” See? What’s old is new again!
By creating valuable, authoritative (read: unique!) content that meets users’ needs, websites can continue to grow their visibility and influence in an AI-driven world. SEO professionals who embrace these changes and integrate AI into their strategies will be better positioned to succeed.
Dispelling the myth of theft
The idea that AI-powered search is built on theft is misleading and overlooks the potential benefits these tools bring. Rather than fearing AI, SEOs should leverage it to enhance their strategies, drive more meaningful engagement, and ensure their content remains at the forefront of search results.
AI isn’t here to steal; it’s here to drive evolution. It’s not the biggest or loudest that thrive on the web, but those who adapt. Just as in nature, survival belongs to the fittest—those who innovate, evolve, and embrace the future.
Carolyn Shelby is an expert in SEO and AI, specializing in enterprise and technical SEO and optimizing web architectures. She views SEO and AI as powerful tools to narrate a brand’s journey, aligning content with core values to engage and convert audiences. Her approach is both data-driven and distinctly human, using straightforward, innovative methods to achieve real results.
September 19, 2024
Team Yoast is at Attending The Global Search awards! Click through to see who will be there, what we will do, and more!
See where you can find us next »
01 October 2024
Get expert analysis on the latest SEO news developments with Carolyn Shelby and Alex Moss. Join our upcoming update! 📺️
All Yoast SEO webinars »
3 Responses to No, AI-powered search is not built on theft: debunking misconceptions
Brooke • 6 days ago
This is info that all users of AI search need to understand, not just SEO specialists!
I’ve had a steady dip (not a dive) in traffic, and I’ve been wondering if it’s related to AI search impact or the Google algorithm change. Since we still rank high for some good keywords and it has happened gradually, I assume it’s a side effect of AI search.
A bot? not a bot? apparantly it doesn't matter anymore • 6 days ago
Except time after time these ‘not thieving’ LLM creators are found to be ignoring robots.txt and website T&Cs to scrape en masse without permission (that would be theft… right???).
Often to the point of effectively DDOSing legitimate websites in order to steal their content.
Of course you want to convince your readers of the boon of AI, given it’s a product you will know doubt want to sell to them. But to claim it isn’t theft is either ignorance, or deliberately trying to cover for criminals no matter how you sugar coat it.
The hilarious thing, Carolyn, is it’s YOUR job the AI will be replacing. Assuming it hasn’t already…
^ Post author
Carolyn Shelby • 4 days ago
I would address you by name, but since you’ve chosen to remain anonymous, I’ll respond to the points you’ve raised directly.
It’s not surprising that The New York Times and others are suing over the use of their content in AI training, just as they did back in the ’90s with deep linking and framing lawsuits. Back then, companies like The Times argued that linking to their content or framing it on other sites without going through the homepage (and bypassing ads) was “theft.” They lost those cases, and the internet kept evolving.
Now, AI is the new “threat,” and this lawsuit is just another attempt to protect an outdated business model. Just like the courts eventually ruled that deep linking wasn’t illegal, the use of publicly available data to train AI models is transformative—LLMs aren’t republishing articles; they’re learning from patterns in the data to generate new content.
What we’re seeing isn’t theft—it’s technological evolution. Just as media companies had to adapt to the internet by developing new business models (like paywalls), they will need to adapt to AI. Instead of calling this progress “theft,” it’s an opportunity to explore ways to innovate and collaborate with AI technologies.
And no, AI isn’t going to replace strategic, high-level work like mine anytime soon. It’s a tool, not a substitute for the creativity, critical thinking, and expertise that professionals bring to the table.
This is info that all users of AI search need to understand, not just SEO specialists!
I’ve had a steady dip (not a dive) in traffic, and I’ve been wondering if it’s related to AI search impact or the Google algorithm change. Since we still rank high for some good keywords and it has happened gradually, I assume it’s a side effect of AI search.
Except time after time these ‘not thieving’ LLM creators are found to be ignoring robots.txt and website T&Cs to scrape en masse without permission (that would be theft… right???).
Often to the point of effectively DDOSing legitimate websites in order to steal their content.
Proof
https://www.theguardian.com/technology/2024/apr/30/us-newspaper-openai-lawsuit
More Proof
https://www.theguardian.com/media/2023/dec/27/new-york-times-openai-microsoft-lawsuit
Of course you want to convince your readers of the boon of AI, given it’s a product you will know doubt want to sell to them. But to claim it isn’t theft is either ignorance, or deliberately trying to cover for criminals no matter how you sugar coat it.
The hilarious thing, Carolyn, is it’s YOUR job the AI will be replacing. Assuming it hasn’t already…
I would address you by name, but since you’ve chosen to remain anonymous, I’ll respond to the points you’ve raised directly.
It’s not surprising that The New York Times and others are suing over the use of their content in AI training, just as they did back in the ’90s with deep linking and framing lawsuits. Back then, companies like The Times argued that linking to their content or framing it on other sites without going through the homepage (and bypassing ads) was “theft.” They lost those cases, and the internet kept evolving.
Now, AI is the new “threat,” and this lawsuit is just another attempt to protect an outdated business model. Just like the courts eventually ruled that deep linking wasn’t illegal, the use of publicly available data to train AI models is transformative—LLMs aren’t republishing articles; they’re learning from patterns in the data to generate new content.
What we’re seeing isn’t theft—it’s technological evolution. Just as media companies had to adapt to the internet by developing new business models (like paywalls), they will need to adapt to AI. Instead of calling this progress “theft,” it’s an opportunity to explore ways to innovate and collaborate with AI technologies.
And no, AI isn’t going to replace strategic, high-level work like mine anytime soon. It’s a tool, not a substitute for the creativity, critical thinking, and expertise that professionals bring to the table.