←back to thread

237 points robin_reala | 3 comments | | HN request time: 0.89s | source
Show context
scq ◴[] No.43514594[source]
This seems like a bug in the ScreenAI service? There's no evidence whatsoever for his claim that Google "trains a machine vision model on the contents of my screen".

According to https://chromium.googlesource.com/chromium/src/+/main/servic... it is just inference.

> These functionalities are entirely on device and do not send any data to network or store on disk.

There is also this description in the Chrome OS source code:

> ScreenAI is a binary to provide AI based models to improve assistive technologies. The binary is written in C++ and is currently used by ReadAnything and PdfOcr services on Chrome OS.

replies(2): >>43514631 #>>43516050 #
bri3d ◴[] No.43514631[source]
This. Go to chrome://flags and disable “Enable OCR For Local Image Search” and I bet the problem goes away.

It’s a stupid feature for Google to enable by default on systems that are generally very low spec and badly made, but it’s not some evil data slurp. One of the most obnoxious things about enshittification is the corrosive effect it seems to have had on technical users’ curiosity: instead of researching and fixing problems, people now seem very prone to jump to “the software is evil and bad” and give up at doing any kind of actual investigation.

replies(8): >>43514714 #>>43514728 #>>43514740 #>>43514895 #>>43514932 #>>43515032 #>>43515378 #>>43515585 #
1. NoNotTheDuo ◴[] No.43514714[source]
> but it’s not some evil data slurp.

Not yet anyway. We’ve just seen Amazon change how all Echo’s/Alexa’s operate. It has been local-only for years and years, but now they want the audio data, so they’ve changed the Terms of Service. There’s no reason to believe Google won’t do the same thing sometime in the future.

replies(1): >>43514947 #
2. simpaticoder ◴[] No.43514947[source]
"Trauma" is when one horrible experience lowers your danger threshold so much that it triggers on everything, and becomes useless and harmful. "Learning" is when new threat awareness lowers the threshold an 'appropriate amount'. Even if the GP was strictly wrong about their conclusion, in my personal opinion they are quite right to remain vigilant.

Note to parent: it is strictly unfair to lump Google in with Amazon (and if you demonize a good actor long enough, eventually they'll aquiesce since they are already paying the reputational price). However given that they are American corporations operating on similar incentives during the Wild West (or World War) of AI aka WWAI, it makes sense to be suspicious. Heaven knows "reputational downside" is just about the only counter-veiling incentive left, since Trump has stripped consumers and investors of virtually all legal protection (see: CFPB elimination; SEC declines Hawk Tua coin grift prosecution; Trump pardons Trevor Milton). I think it is an excellent time for all of us to be extremely careful with the software we use.

replies(1): >>43515036 #
3. broknbottle ◴[] No.43515036[source]
Google is an Advertisement company. Everything they do revolves around slurping up the most valuable data to better identify people and be able to identify trends. They’ve become increasingly less and less open as year goes by and they still haven’t found their next big cash cow to offset decline to their current cash cow.