

Reading this, all I can think about is all the confidential information that are probably getting leaked to the AI providers.


Reading this, all I can think about is all the confidential information that are probably getting leaked to the AI providers.


According to the article, they use the gained time to do more work. This doesn’t seem like a win to me, they do more work but still are paid the same. In fact, if they pay for AI, then they get even less while providing more value to the corporation.


Neet, if NixOS get used more and more then the skills I acquired making my config will become marketable.


That’s a totally valid use case, I have done it in the past and there are Linux distributions that are made to be used this way. Those distributions usually work by loading the whole system into ram which avoid multiples writes to the USB stick. They also usually offer some persistence mechanism so that data can be added on the USB stick.
Examples of distribution meant to be used this way: MX Linux, Puppy Linux


Sharing this because I think it’s funny, but someone opened an issue on Mullvad’s github about the questionable donations of one of the cofounders: https://github.com/mullvad/mullvadvpn-app/issues/10685


This honestly seem pretty good, they even offer mail aliasing. I think I’m gonna migrate to them once my one year subscription to Mullvad end.
I genuinely cannot believe that they think this is a good idea. After their whole Sora thing failed, you would think they would have gotten wiser, guess that’s not the case. I cannot see a world where this thing do anything, but flop. Maybe they will sell them at a lost, but, even then, I have doubt that it will succeed.


Interesting, I didn’t consider they could do it this way, but it make perfect sense. Thanks for bringing this up.


I wasn’t understanding what use keystrokes and mouse-clicks would have without also capturing what’s happening on screen. Turn out, they are also capturing that:
The tool, called Model Capability Initiative (MCI), will run on work-related apps and websites and will also take occasional snapshots of the content on employees’ screens
This is still pretty much bleeding edge technology, who’s to say that the way we are currently doing thing is the most optional one? I’d rather they stop building these massive data centers and instead have their researchers focus on making these models run on consumer level hardware since throwing more and more compute power at the problem is not sustainable in the long run.