How Search Engines Are Reevaluating Trust in Response to Generative AI
Search engines have always tried to help users complete tasks with the most accurate and credible information, and that has not changed with the arrival of generative AI. What has changed is the volume and sophistication of the content competing for users' trust, as well as the emergence of a new interface that major search engines like Google and Yandex use to provide answers.
AI-generated answers are built on top of the same underlying search and retrieval work that has always determined which sources are worth surfacing. They do not replace that evaluation. Instead, they add a new layer to it and raise the bar for what counts as a comprehensive and trustworthy answer. Brands need to understand that distinction to stay visible in AI-powered search.
How AI Changed the Search Landscape
AI-powered search systems can now scan enormous volumes of content before synthesizing a response to the user’s query. This capability marks a real shift in the search landscape and in how search functions. For a long time, the main challenge in search was identifying the pages most relevant to a user’s query. The new challenge is figuring out which of many relevant sources can provide the best answer.
The internet has become as competitive as almost any offline market — arguably more so — and nearly every topic is now covered by dozens of credible-looking sources. The rise of AI-generated content has also made it much easier to create and publish articles that look credible but lack the depth and accuracy that users need. Generative search systems do not eliminate that competition, but they are now designed to select the handful of sources that are most trustworthy when building the answer a user sees.
What Determines Trustworthiness
Trustworthiness is not a direct signal that generative search systems can measure. Instead, it can be inferred from the quality of your content and its ability to actually address the task behind each user query.
The EPOS Framework
Yandex’s EPOS framework provides a useful way to think about content quality. The acronym stands for Expertise, Practicality, Originality, and Substance:
Expertise: whether the content has genuine depth and judgment behind it, the kind that comes from someone actively working in the field.
Practicality: whether the content helps the user complete a task or make a decision, rather than merely describing relevant topics in the abstract.
Originality: whether the content adds original information or data not found elsewhere on the internet.
Substance: whether the content goes deep enough to be genuinely useful to the user.
These four qualities describe what search engines have always tried to identify: the source that best addresses the user’s problem. Content that is useful, original, and sufficiently detailed is much more likely to be used as a source by AI-powered search engines like Google, Yandex, and Bing.
Understanding Query Intent
Another major part of adapting to generative AI search is focusing on user intent rather than exact keyword matches. When creating their content strategies, it is more useful for brands to ask what the person is trying to achieve, what stage of the decision-making process they are at, and what related question they are likely to ask next.
Systems like Yandex’s Alice AI may expand a question into related subtopics and draw on the sources that cover them most effectively. If that expansion pattern holds across other generative platforms, it may be better to build pages that anticipate a variety of related user questions than to optimize a page for a single target query.
Content Depth over Broad Coverage
To comprehensively answer user questions and address the tasks behind their queries, content needs depth. Covering a wide range of topics can result in thin, less useful content. A page that thoroughly resolves the problem underlying a user’s question and anticipates several related questions is a much stronger candidate for citation.
Broader External Validation
Building relationships with other credible sources remains an important signal of trustworthiness. Brands should take a broader view of external validation to include comparison pieces, independent reviews, forum discussions, and other sources of third-party commentary.
It is difficult for a company to publish a credible head-to-head comparison involving itself on its own website, but a trade publication or an industry analyst can do just that. A business’s overall information footprint will help generative AI systems decide which sources are most trustworthy.
Key Takeaways
Generative AI is reshaping how people find and receive information, and that is changing how brands should approach visibility. However, trust is not a direct signal that can be improved through a series of technical adjustments. The brands that continue to be trusted by AI-powered search engines are likely to be those that focus on solving users’ problems and treat expertise, practicality, originality, and substance as goals in themselves.
