31–40 of 203 posts
Privacy aside if it's done server side, I think this crosses another dangerous line. With this, Google controls how content you read is summarized, they could introduce biases, intentionally or not, and I'm not sure we should trust such an actor for this.
This was always the case.
I assume this is all done server-side, with Google feeding your interests and investigations into your personal profile. I really wish this could be done on-device. I don't want Google slurping up even more data about web users. I use FF mainly on desktop and mobile, and there's probably a great opportunity for Mozilla to build an offline, privacy-first summarization model.
I’m wildly speculating, but it seems that projects like llama.cpp are bringing SOTA models closer to the desktop. It’s only a matter of time before browsers embed a small LLM for various purposes, providing access to this local model via a JavaScript API that allows clients like Gmail to perform tasks on data locally. Apple would be strongly incentivized to do this, given their value proposition of user privacy.
Build as much of the model as you can in the cloud, run inference locally and push results back is probably the cost optimal way to run this stuff at scale.
Privacy aside if it's done server side, I think this crosses another dangerous line. With this, Google controls how content you read is summarized, they could introduce biases, intentionally or not, and I'm not sure we should trust such an actor for this.
Google has always introduced biases by virtue of ranking your results and deciding which content to surface. This was always the case.
I assume this is all done server-side, with Google feeding your interests and investigations into your personal profile. I really wish this could be done on-device. I don't want Google slurping up even more data about web users. I use FF mainly on desktop and mobile, and there's probably a great opportunity for Mozilla to build an offline, privacy-first summarization model.
[0] https://help.kagi.com/kagi/getting-started/faqs.html#i-have-...
For the record I am a happy user, not affiliated.
I assume this is all done server-side, with Google feeding your interests and investigations into your personal profile. I really wish this could be done on-device. I don't want Google slurping up even more data about web users. I use FF mainly on desktop and mobile, and there's probably a great opportunity for Mozilla to build an offline, privacy-first summarization model.
I assume so too, and given that, it's incredibly frustrating (but not at all surprising) that they require users to use Chrome to be able to use the feature.
A few hours ago I wanted to know the difference between a typhoon and a hurricane. Here is the answer: they're the same thing, just in a different ocean. I clicked three or four articles and had to read for a few minutes to get to that answer.
It's even worse with videos. It takes 10 minutes to answer the simplest questions, because that's the ideal length for monetisation.
Articles are getting too long? Since when has this been a problem? I think it's the amount of crap ads injected while I'm trying to read the article that is the main problem.
It's not actually a direct correlation, but enough people take that as gospel that they will pad out an article to make it super long just for the SEO value.
Similar to recipe websites giving their entire life story before a recipe.
Earlier quoted context omitted.
Solution: dont use google.
Seems like that's the direction things are moving. Google's becoming heavily enshitified.
Their entire business was built on this. https://en.m.wikipedia.org/wiki/List_of_mergers_and_acquisit...
I know this is HN and all, but to wait all of these years of acquisitions, poaching, project terminations, Chrome shenanigans to destroy the internet, main product "SEO" degradation and literally every single sign of enshittification...Google has always been one of the prime examples of enshittification. That's why they are very successful at making money.