Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI overviews cause massive drop in search clicks

arstechnica.com

Post reply on HN
Print view View original

421–430 of 872 posts

Re: AI overviews cause massive drop in search clicks

#421

Post by blendergeek »

hereonout2 wrote: post #347
littlecranky67 wrote: post #233

Here is the experience when clicking a link on mobile: * Page loads, immediately when I start scrolling and reading a popup trying to get tracking consent * If I am lucky, there is a "necessary only". When unlucky I need to click "manage options" and first see how to reject all tracking * There is a sticky banner on top/bottom taking 20-30% of my screen upselling me a subscription or asking me to install their app. U…

You forgot the part about when you actually get to the content, there's usually about 5 paragraphs of SEO filler text before it actually gets onto answering the topic of the post.

or even worse, the page itself is just an AI summary of the topic

Re: AI overviews cause massive drop in search clicks

#422

Post by gorbypark »

zdragnar wrote: post #105
esafak wrote: post #67

Earlier quoted context omitted.

Write documentation and don't block crawlers.

There's a library I use with extensive documentation- every method, parameter, event, configuration option conceivable is documented. Every so often I get lost in the docs trying to do something that actually isn't supported (the library has some glaring oversights) and I'll search on Google to see if anyone else came up with a similar problem and solution on a forum or something. Instead of telling me "that isn't su…

There's an attempt to kinda have these things documented for AIs, called llms.txt, which are generally hosted on the web.

In theory, an AI should be able to fetch the llms.txt for every library and have an actual authoritative source of documentation for the given library.

This doesn't work that great right now, because not everyone is on board, but if we had llms.txt actually embedded in software libraries...it could be a game changer.

I noticed Claude Code semi regularly will start parsing actual library code in node_modules when it gets stuck. It will start by inventing methods it thinks should exist, then the typescript check step fails, and it searches the web for docs, if that fails it will actually go into the type definition for the library in node_modules and start looking in there. If we had node_modules//llms.txt (or the equivalent for other package managers in other languages) as a standard it could be pretty powerful I think. It could also be handled at the registry level, but I kind of like the idea of it being shipped (and thus easily versioned) in the library itself.

Re: AI overviews cause massive drop in search clicks

#423

Post by AlecSchueler »

graemep wrote: post #413
OtherShrezzing wrote: post #222

Earlier quoted context omitted.

In the UK we've got amazing National Health Service informational websites[1], and regional variations of those [2]. For some issues, you might get different advice in the Scottish one than the UK-wide one. So, if you've gone into labour somewhere in the remote Highlands and Islands, you'll get different advice than if you lived in Central London, where there's a delivery room within a 30 minute drive. Google's AI ov…

> For some issues, you might get different advice in the Scottish one than the UK-wide one its not a UK wide one. The home page says "NHS Website for England". I seem to remember the Scottish one had privacy issues with Google tracking embedded, BTW. > So, if you've gone into labour somewhere in the remote Highlands and Islands, you'll get different advice than if you lived in Central London, where there's a delivery…

> But someone in a remote part of England will get the same advice as someone in central London,

The current system might not have perfect geographic granularity but that doesn't mean it isn't preferable to one that gives advice from half the world away.

> Human biology is the same, diseases are the same, and the difference in available treatments is not usually all that different

Accepted medical definitions differ, accepted treatments differ, financial considerations, wait times and general expectations of service vary wildly.

Re: AI overviews cause massive drop in search clicks

#424

Post by nosianu »

cornholio wrote: post #167
throwaway81523 wrote: post #133

Of course slow, shitty web sites also cause a massive drop in clicks, as soon as an alternative to clicking emerges. It's just like on HN, if I see an interesting title and want to know what the article is about, I can wince and click the article link, but it's much faster and easier to click the HN comments link and infer the info I want from the comments. That difference is almost entirely from the crappy overdesig…

This is a pretty apt analogy: why settle for the original article when you can read the outrage infused summary of an opinionated troll in a hurry? It has little to do with overdesign or load times.

The majority of the linked articles is waaayyyyy too long for what they have to say, and they reveal the subject only many paragraphs in.

From reading one or a few short comments I at least know what the linked article is about, which the original headline often does not reveal (no fault of those authors, their blogs are often specialized and anyone finding the article there has much more context compared to finding the same headline here on a general aggregation site).

Re: AI overviews cause massive drop in search clicks

#425

Post by svachalek »

azangru wrote: post #377
jdietrich wrote: post #315

Earlier quoted context omitted.

It's a market where nobody has a particularly deep moat and most players are charging money for a service. Open weight models aren't too far behind proprietary models, particularly for mundane queries. The cost of inference is plummeting and it's already possible to run very good models at pennies per megatoken. I think it's unreasonably pessimistic to assume that dark patterns are an inevitability.

> and most players are charging money for a service The aricle talks about AI overviews. As exemplified by the AI summary at the top of Google search results page. That thing is free.

1. Create free and good product

2. Attract large user base

3. Sell user data and attention to advertisers

4. Extract maximal profit from sponsors

5. Earn billions from shit product

Re: AI overviews cause massive drop in search clicks

#426

Post by cudder »

reddalo wrote: post #367
cudder wrote: post #366

Earlier quoted context omitted.

There is an extension called "I still don't care about cookies" that mostly solves the nag screens (There's also a similar one that doesn't have the "still" in its name but that one was bought by an ad company and enshittified.) AFAIU it usually accepts the cookies though, so you should combine it with something that clears your cookies periodically. Sometimes it breaks the site so that you can't scroll or something,…

No need to use a specific extension, uBlock Origin is all you need. Just enable the "Easy List/uBO - Cookie Notices" filter from the filters list (the default is off).

Sweet, didn't know that. I'm gonna try if it works with less breakage. Thanks!

Re: AI overviews cause massive drop in search clicks

#427

Post by seunosewa »

moontear wrote: post #365
generic92034 wrote: post #361

Earlier quoted context omitted.

Competition or not, dark patterns or not - sooner or later LLMs will need to earn money for their corporations.

But they do? Paid subscriptions for Gemini, ChatGPT and Copilot are a thing. If Google throws in a free AI summary in their search it only helps promoting Gemini in the long run.

The mediocre AI summaries aren't promoting Gemini when you can't use them to start a chat on Gemini. They effectively ads and search results for no benefit.

Re: AI overviews cause massive drop in search clicks

#428

Post by Disposal8433 »

hereonout2 wrote: post #347
littlecranky67 wrote: post #233

Here is the experience when clicking a link on mobile: * Page loads, immediately when I start scrolling and reading a popup trying to get tracking consent * If I am lucky, there is a "necessary only". When unlucky I need to click "manage options" and first see how to reject all tracking * There is a sticky banner on top/bottom taking 20-30% of my screen upselling me a subscription or asking me to install their app. U…

You forgot the part about when you actually get to the content, there's usually about 5 paragraphs of SEO filler text before it actually gets onto answering the topic of the post.

I have noticed that a lot. For example:

What is the price of the Switch 2?

The Switch 2 can be purchased with money.

Re: AI overviews cause massive drop in search clicks

#429

Post by svachalek »

littlecranky67 wrote: post #300
tmountain wrote: post #279

Earlier quoted context omitted.

AI is following the drug dealer model. “The first dose is free!” Given the cost incurred, lots of dark patterns will be coming for sure.

I fail to see how that will work out. Just I have an adblocker now, I could have a very simple local llm in my browser that modifies the search-AIs answer and strips obvious ads.

They won't be obvious. They'll be highly customized brain worms influencing your votes and purchases to the highest bidder.

Re: AI overviews cause massive drop in search clicks

#430

Post by AlecSchueler »

gorbypark wrote: post #422
zdragnar wrote: post #105

Earlier quoted context omitted.

There's a library I use with extensive documentation- every method, parameter, event, configuration option conceivable is documented. Every so often I get lost in the docs trying to do something that actually isn't supported (the library has some glaring oversights) and I'll search on Google to see if anyone else came up with a similar problem and solution on a forum or something. Instead of telling me "that isn't su…

There's an attempt to kinda have these things documented for AIs, called llms.txt, which are generally hosted on the web. In theory, an AI should be able to fetch the llms.txt for every library and have an actual authoritative source of documentation for the given library. This doesn't work that great right now, because not everyone is on board, but if we had llms.txt actually embedded in software libraries...it coul…

> In theory, an AI should be able to fetch the llms.txt for every library and have an actual authoritative source of documentation for the given library.

But isn't the entire selling point of the LLM than you can communicate with it in natural language and it can learn your API by reading the human docs?

Post reply on HN