How AI is Killing the Search Market

The Disruptive Move from Search Engines to Answer Engines

When you search on Google today, you could do what you have done for years. You could peruse the 10 blue links on the first page and even more. What you would then typically do is integrate the findings to arrive at a conclusion. During this process, you might easily click on one of the Google Ads- a seemingly innocuous act that will result in advertising revenue for Google and traffic for the advertiser.

All this has changed.

First, what you get is an AI-generated summary. You don’t have to travel to various websites and try to discern what each has to say. Google’s Gemini does that for you. It visits multiple sites and synthesizes the findings. We know that generative AI is very strong in summarizing content- so, this plays to the core of the technology.

But, wait. There’s more.

Secondly, Google has introduced something called AI Mode. The AI mode fundamentally changes user behavior. Rather than asking a typical search engine query (most likely to be a short phrase of 4-5 words), it encourages users to prompt Google in AI Mode.

Business Model Impact

These changes have phenomenal implications for the business model underlying the Internet.

Change in website traffic

The whole art of search engine optimization was about placing your link at the very top of the heap. Organizations routinely hired SEO experts who played with Google’s PageRank algorithm to game the number of a backlinks to a page and added meta tags etc. All of that will go to naught if the user is in AI mode and is not going to the advertiser’s site.

In the world of answer engines, Google alone goes to an individual website. The user doesn’t. Therefore, it is no wonder that major publishers (e.g., Washington Post, Business Insider) who rely on organic traffic are showing massive declines in traffic (Source: WSJ).

The only publisher whose traffic is unaffected is the Wall Street Journal- which relies on greater subscription revenue rather than advertising.

Cannibalization of Advertising Revenues

AI Overview and AI Mode are fundamentally antithetical to Google’s current business model. Google understands this. But, it is betting that AI Overview and AI Mode are habituating consumers to the idea of Google as an answer engine. Over time, they could move advertisements to the AI side. For instance, certain sites that pay more could receive greater sway in the way Google displays AI-generated results.

Google understands that, in the long run, this changes the equation between the publisher and the advertiser. If anything, it positions Google as a stronger authority- i.e., we will move page authority (as measured by PageRank) to platform authority (i.e., Google as the trusted provider of answers).

Perplexity is the Competitor

For the first time in decades, Google has an existential thread- Perplexity. It positioned itself as an answer engine from inception leading to impressive market evaluation. Given the clarity of the positioning and the impressive product, Google has been forced to innovate and respond. In the meantime, there are rumors that Apple is in serious talks to acquire Perplexity.

Societal Impact

Ethical Implications of AI

This is best summarized in this paper-

First, LLMs are known to hallucinate information and cannot detect factual inconsistencies (Venkit et al., 2024; Huang et al., 2023), even when authoritative sources are provided. Second, prior work (Liu et al., 2023; Laban et al., 2024) has also shown limitations of answer engines’ ability to assess the accuracy of citations within an answer. Third, LLMs accumulate knowledge within their internal weights during pre-training, and prior work has shown limited success at enforcing that an LLM generates information based solely on documents provided in a prompt rather than based on pretraining information which can be noisy or outdated (Kaur et al., 2024). Finally, such systems exhibit sycophantic behavior: a preference for the system to agree with the user’s perceived opinion over objective truth (Sharma et al., 2024; Laban et al., 2023b).

Put otherwise, we have built an information economy with the basis of factual information from Google. We are now redefining our digital social contract through the creation of answer engines.

Everybody will read less

The same paper found that the traditional researcher hovered and clicked on more citations in search engines as opposed an answer engine.

Impact on Education

This will require further reflection and discussion.

In short, with AI Mode, the critical faculties of the students are less likely to be taxed with easy answers to questions being served up. As educators, we must find innovative ways to advance our educational methods to respond to these changes. Recombinant AI fluency might be the way forward.

Conclusion

Googling has been part of our lexicon for a few decades now. We are used to searching for information, sifting through the results, curating the ones that matter and then integrating findings across different weblinks.

AI now performs those tasks at machine scale. The open question is whether its synthesis mirrors human judgment — or substitutes it. Organizations that depend on search traffic, ad impressions, or even student inquiry must rethink strategy.

How will your business adapt when the click-through becomes optional?