New AI Search Engines Are Citing the Wrong Sites - Here's Why.
Artificial intelligence-powered search engines are often more likely to cite lesser-known websites, sources that wouldn't even appear in Google's top 100 links. Researchers from Ruhr University and Max Planck Institute for Software Systems studied how these search engines worked by comparing their results to traditional search engine links.
Their research found that AI search engines tend to cite sources that are less popular than those listed in the top 10 of a traditional search, with some sources falling outside the top 1,000 domains tracked. One search tool, Gemini, was particularly prone to citing low-popularity domains.
While AI-powered search results may not provide as much detail or diversity as traditional links, they tend to cover similar concepts and are often more concise. However, these engines can become less effective when searching for timely information, sometimes responding with generic requests for more information rather than conducting a web search.
Researchers emphasize the need for new evaluation methods that consider source diversity, conceptual coverage, and synthesis behavior in generative search systems. The question remains whether AI-based search engines are overall better or worse than traditional search engine links.
Artificial intelligence-powered search engines are often more likely to cite lesser-known websites, sources that wouldn't even appear in Google's top 100 links. Researchers from Ruhr University and Max Planck Institute for Software Systems studied how these search engines worked by comparing their results to traditional search engine links.
Their research found that AI search engines tend to cite sources that are less popular than those listed in the top 10 of a traditional search, with some sources falling outside the top 1,000 domains tracked. One search tool, Gemini, was particularly prone to citing low-popularity domains.
While AI-powered search results may not provide as much detail or diversity as traditional links, they tend to cover similar concepts and are often more concise. However, these engines can become less effective when searching for timely information, sometimes responding with generic requests for more information rather than conducting a web search.
Researchers emphasize the need for new evaluation methods that consider source diversity, conceptual coverage, and synthesis behavior in generative search systems. The question remains whether AI-based search engines are overall better or worse than traditional search engine links.
! Like, I know it's cool to cite lesser-known websites and all, but what if you're actually looking for some legit info? 
It sounds like these AI engines are just spitting out random sites that aren't even in the top 100. And Gemini is just straight-up problematic with how it cites low-popularity domains 
Maybe this means we'll finally see some fresh perspectives and less repetition in our search results? 
On the flip side, though, if these engines can't even handle timely searches, like, what's the point? 
It seems to me that we need some new evaluation methods ASAP, 'cause right now it's all just a big mystery 
Like, what does that even mean? Are those legit sources or just some dodgy sites we don't want to look at again?
. Can't we just find a sweet spot between the two? Maybe some researchers could look into that next?
... but I think the problem is, they're not always reliable 
. Like, who needs a generic answer from an AI if you just want some info on what's trending right now? 
It'd be great if these search engines could get better at handling timely queries and not just return basic requests for more information... that's like, super frustrating 
. And it's also kinda weird when they cite sources that are way outside the top 100 or even 1,000 
. I mean, I get it, they might be covering similar concepts and all that jazz, but when am I supposed to find a relevant article about current events? Those generic requests for more info are just frustrating 
. It's like they're trying to be all unique and stuff, but really they're just missing the point 
. Can't we just stick with Google for now? 
, which could be useful for finding unique info. But on the other hand, it's weird that they can't provide as much detail or diversity as traditional search engines, which is kinda what makes 'em awesome in the first place 
. Overall, I think AI search engines have potential, but we need to figure out how to make 'em better, ya know? 
and it's wild to think that those low-popularity domains are getting mentioned alongside legit sources. it's like they're trying to throw us off track 
. but, on the bright side, ai search results can be way more concise and cover similar concepts as traditional links 
It's interesting that these AI engines might provide more concise answers, but sometimes I feel like they just don't get the info right away. Like when we used to ask Siri or Alexa to give us directions, now it's all about asking them for more info... what's up with that? Maybe we need new rules for evaluating these search tools or something? 
... I mean, what if we start relying on them too much? I've tried Gemini and it's kinda cool how concise the results are, but sometimes it feels like it just regurgitates stuff without even checking for accuracy 
But seriously, it's kinda mind-blowing how different they can be from the top results on Google. Like, I was searching for something recently and this AI thingy gave me a link that didn't even appear on my friend's Google search 
.
I mean, can you believe AI-powered search engines are citing those super obscure sites? That's wild