made you look
To me lying implies an intent to deceive, LLMs can’t do that as they have no intentions or understanding of the output they produce.
It’s not lying, because it’s also not telling the truth either, it’s just statistically weighted noise.
I kinda get the average person overestimating the capabilities of an LLM, but technically minded people falling for it did surprise me.
Of course, I really shouldn’t be, the ELIZA effect is old and well documented, I suppose it’s more worrying that it can affect anybody.
Nah, storage is fried.
People always focus on systemd whenever this is posted, but all systemd is saying is that it can’t read the service files when it tries to start something. Earlier on the kernel is complaining about I/O errors as well.
That was Ars Technica.
I bet the actual logo display is a full screen browser too, multiple computers each running chrome just to display ads.
By claiming that you own patents on technology used by said format.
The “open royalty free” aspect applies to companies that are a part of the AOMedia group, if you’re not involved with them you’re not covered by the patent grants and restrictions in place, and can charge whatever the courts say is cool.
Depending on the output device it’s still using ALSA underneath (e.g. Bluetooth output instead is given to the BT stack), PipeWire is dealing with managing and routing the audio output rather than actually performing it.
The best part of the article is the very end, even if the site makes it look unrelated.
Avanci’s Video pool and Access Advance’s Video Distribution Patent pool are both now seeking content royalties from streaming services for the use of HEVC, VVC, VP9, and AV1. Access Advance’s rates are capped at roughly $63 million per year, and Avanci has published rates of 1.6% to 2.0% of revenue or $0.12 to $0.15 per user per month.
$4.5 million max for H.264 is rookie numbers vs. the $63 million max for AV1
There were also compatibility issues with the “CSS box model”, where IE6 didn’t follow the spec at all and broke nearly every site in every other browser because elements ended up with different sizes.
They fixed that with IE7, and we finally entered the utopia CSS promised with every browser agreeing on how to size elements.
And then everybody realised the CSS defaults were wrong, and the IE6 behaviour actually made more sense, and now pretty much any complicated site will opt back into the IE6 box model.
Yeah, we have mDNS for a reason.
Or even just link the DHCPv6 server to the DNS, that’s the default config in most cases anyway.
tbf it was played straight in Hot Fuzz, Sgt. Angel was right and there was somebody who caused it.
True, but it’s not unheard of. FreeBSD did the same thing in their latest release.
Which makes sense, IPv6 is actually quite backwards compatible with IPv4, but v4 isn’t forward compatible in the same way.
Edit: It’s in the release notes for FreeBSD, under Networking. Now possible to disable IPv4 while keeping IPv6 enabled, and build a single stack kernel.
Makes it portable across architectures while also providing sandboxing.
The fedi software I use (GoToSocial) runs both ffmpeg (Sorry, ffmpreg) and sqlite through WASM, also makes it easier to integrate it with Go code apparently.
There’s BlackSky now, the first full outside server setup (Things like relays and PDSs are just smaller components of the larger required stack)
So you know, they’re at 2 total instances currently.
I remember seeing that years ago, wanted to make like a photoresist mask to etch it into metal.
These days you could probably feed it to a laser engraver, get some nice depth on a thicker sheet of e.g. aluminium, would be a nice display piece at least.
It’s not just bug reports; in the last month, AI driven development has actually gone from slop to reliably better than the average human.
Funny, I heard that same claim about 6 months ago.
And I’m sure I’ll hear it again in another 6 months or so.
will hopefully drive positive govt policy outcomes.
From the current city and state governments? Highly unlikely.
The laws they’re trying to support require very limited information, but they’re storing far more than that and they’ve actively decided not to protect it properly.
All systemd is storing is the DOB in YYYY-MM-DD format.
Ok, but who is making those “open weight” models though? Individuals don’t really have the resources to run these huge scraping operations, so they’re often still corporate releases with fake open source branding.