This will solve itself. If clicks are worth less, companies will pay less for them. It's a bidding system
> but the advertiser may still be charged.
Oh, no. Anyhoo...
But in all seriousness, ads are one of the main reasons quality has decreased online. If AI breaks the ad-driven economy and forces the internet to be something else, ya know... maybe that is not so bad.
The idea came from noticing that AI agents performing web research could accidentally generate ad-click costs at scale.
If millions of automated queries start interacting with ad ecosystems, even small unintended click rates could accumulate.
Curious whether agent frameworks should treat sponsored links similar to robots.txt — something to avoid unless explicitly allowed.
But isn't there already click distortion from web scrapers?
Yes, sites should support a NoAds header that agents can provide, which ensures that the site doesn't provide any ads that the agent could accidentally click on.
The site would then be providing information to the AI, without and means of generating revenue to keep going. Where is the incentive for commercial sites to play this game?
This came up years ago with the Google Answer Box. People were pretty upset that Google wasn’t sending people to the actual websites anymore and just scraping the relevant content. It was seen stealing potential ad revenue from the author who did the work to put the information out on the internet.
I haven’t heard a lot of talk about out about this with LLM, even though they are like the Google Answer Box on steroids. What I’ve heard more is the general talk of IP theft during training.