And then when those AI also have issues do we use the AI to check the AI for the AI?

I assume a lot of android foss app developers are going to refuse to register and the projects are going to need to be forked.

Personally I’m getting an old feature phone and an ipad mini that only has wifi. If my choice is between apple iOS and google iOS I’d rather just not use anything to do with Google.

Unless revanced somehow gets the APKs resigned it will almost certainly destroy revanced.

iOS has had this same system for forever and nobody’s ever (seriously) claimed you could sideload on there.

If you’re putting in that much work, please submit those edits to musicbrainz! We need all the help we can get 😭

As a musicbrainz editor, don’t depend entirely on Picard and musicbrainz for correct tagging either cause shit isn’t as well curated as you think.

You need to encode the metadata in a standard way, encode new data that shows up in a standard way, and various people can add more metadata to files: think like Posix ACLs or the immutable flag or whatever.

Nix actually invented a fork of tar specifically for this called “normalized archive” or “Nix Archive” or nar. Guix uses this too:

https://releases.nixos.org/nix/nix-2.22.0/manual/protocols/nix-archive.html

“here is a checksum of my /usr/lib folder, if yours differs from mine I can’t provide installation support”

You won’t find anything as robust as the database of reviews that Google maps has built up just due to their size and ubiquity, you’ll have to see if you can find reviews on non-maps review sites like tripadvisor, yelp, OpenTable etc.

That’s good and also somewhat disappointing as they were the first to release the weights and mechanism to run them as open weights.

A lot of fully open source (and “ethically trained”, depending on your opinion of that entire idea) models still use major portions of the code they open sourced.

A lot of relatively “good” LLM models run on top of Llama.cpp

Hey, Guix user here, I…if you want we… We could recommend to each other…

Payment processors are monopolies for online and electronic payments because there is no global infrastructure (outside Chinese offerings) for electronic payments that don’t rely on Visa and Mastercard.

Something can be Libre without it’s initial distribution being free (as in beer).

That’s how many distributions used to do it in the days when CDs and floppies etc weren’t exactly free.

“it says here you clicked ‘sign me up for ISIS’ 10000 times?” “Haha no officer, you see it was my social chaff AI that clicked it”

Is this just hosted nextcloud with collabora office pre installed?

Shepherd my love!

I’ve used SystemD for years and the pure joy writing system initialization units in Scheme gives me can’t be overstated.

Seriously, a lot of times I feel like I stick with Guix’s many problems just for shepherd.

I have also been done in many times by git-filter-repo. My condolences to the chef.

Outdated image, everything goes through palantir now

There’s a lot of assumptions about the reliability of the LLMs to get better over time laced into that…

But so far they have gotten steadily better, so I suppose there’s enough fuel for optimists to extrapolate that out into a positive outlook.

I’m very pessimistic about these technologies and I feel like we’re at the top of the sigma curve for “improvements,” so I don’t see LLM tools getting substantially better than this at analyzing code.

If that’s the case I don’t feel like having hundreds and hundreds of false security reports creates the mental arena that allows for researchers to actually spot the non-false report among all the slop.

It found it 8/100 times when the researcher gave it only the code paths he already knew contained the exploit. Essentially the garden path.

The test with the actual full suite of commands passed in the context only found it 1/100 times and we didn’t get any info on the number of false positives they had to wade through to find it.

This is also assuming you can automatically and reliably filter out false negatives.

He even says the ratio is too high in the blog post:

That is quite cool as it means that had I used o3 to find and fix the original vulnerability I would have, in theory, done a better job than without it. I say ‘in theory’ because right now the false positive to true positive ratio is probably too high to definitely say I would have gone through each report from o3 with the diligence required to spot its solution.