Search SystemsSearch Evaluators

[best chromebook]

https://twitter.com/AravSrinivas/status/1792689564357914704date: 2024-05-20
systems: Google, Perplexity AI

key phrases: "navigational query"
evaluating: sponsored search results, unclear focus, SERP
[...] The question is whether it benefits the user or the advertiser and whose benefit ultimately benefits Google more. [...]
Aravind SrinivasCEO at Perplexity AI

I watched Nilay of Verge's interview with Sundar. Nilay asked him a simple question: why are we not getting AI overviews when I type in "best Chromebook" and instead get many links/ads? the answer was "we only fire it when we think it's beneficial". The question is whether it benefits the user or the advertiser and whose benefit ultimately benefits Google more. A lot of people texted me saying folks like JCal or Sacks declared, "We are screwed, and it's over for us." Reality is different. People are soon going to realize that a product should be intended for one clear single purpose for the user and work without variability and guessing what's going to happen. This applies the other way, too: you don't need AI summaries for every single navigational query that has already annoyed users. There is a sweet spot between LLM summaries, links, answer boxes, widgets, graphs, cards, panels, etc - what exactly it is - is unclear - hence why there isn't a clear winner in this race yet.

Resources: