Moving to AI-powered search and the opportunity for obfuscation.
Dan Davies writes a lot of interesting stuff. You should check out his Substack. In particular, this recent post had some interesting thoughts on Google’s move to its AI-powered search. His big takeaway is that, from the perspective of ad sales, not much will change. There may be movement of ad revenue from the pages Google use to send you back to Google, but overall the impact would be minimal.
As an alternative perspective, consider how an AI-driven search bar differs from the traditional Google results page. Search results are exactly that, multiple results, ordered based on Google’s search algorithm, prioritization, etc. In particular, sponsored results are paid for and presented in a place of prominence on the site.
An AI-powered result is fundamentally different. It is expected to be the answer, such that you don’t need to scroll through results. The other major difference is the format. It’s free-text, with all of the AI-generated text trappings: markdown-formatted lists, multiple alternatives, and over-eager prose.
Consider what Davies said in his post:
Being able to carry out a semantic search - something like “find me three examples of trade associations making outlandish claims about the economic impact of housing regulation” is a real time saver, and although you do have to do a lot of back-and-forth with the AI to get what you want, it’s less of a pain in the arse than filling your screen with browser tabs and reading the press releases yourself.
This may be true from a research-for-knowledge perspective (as in, I want to write an article about this or learn more), but what about a consumer? Say I want the top three laptops for my kid who wants to learn how to program? How is “best” being defined in the generated text? And (here is the big item) what happens when Google realizes that they can charge to get products into that generated response? Each refinement of your query leads to a new output, with possibly new recommendations, each with a new opportunity for Google to collect revenue from firms who want their product included in that generated output. Let’s call these responses semantic salesmen. They’ll iterate to get results, but at the same time will pitch products that might not exactly align with what the consumer wants, so that they can get the deal (ad revenue) closed.
Consumers might not like this. But they didn’t like sponsored search when it rolled out either and (for the most part) people got used to it. Complaints were raised, design modifications were made, and the sponsored results remained. I’d say we’ll see a similar pattern here.