Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror

Comment AI slop = 10+yrs/4+yrs Diamond league and I'm out (Score 5, Interesting) 24

10 years using Duolingo, 4 years in Diamond League, and earlier this month I just had enough of the low-quality AI bullshit.

Out. Done. Gone. Waste of my time, every damn day.

The decline in quality and accuracy was just too much. As an English speaker learning a handful of other languages and concentrating on two, it was interesting to me that the first real noticeable patterns of errors and general slop were actually on the English side. Increasingly obtuse questions or statements, ok fine, I will spit those back in French or Spanish with good accuracy. I don't have any issue whatever with the occasionally-dark humor or the gender related topics that might push others' hot buttons, and I appreciate the occasional foray into curious stories and situations. But... over the past year there has been an increasing level of nonsensical AI-generated questions, erroneous answers accepted, multiple correct answers, etc etc... and it's obvious that no actual native speaker looked at a lot of the new content -- either from the native or foreign perspective. A couple years ago there was an increasing level of having to hit the button for "You should have accepted my answer." But for the past year, it's become a daily occurrence to have to hit the button for "You shouldn't have accepted my answer." The latter is a clear indication of AI slop and drift in the language models, and lack of QA. Real human QA is not optional, and Duo has apparently dispensed with it entirely. The result is gameified garbled nonsense. Playing the game was fun for a while (seriously, still in Diamond league for more than four years straight), but the goal is language learning not to compete with other stupid little games on my device. Feh. Done. Cancelled my subscription, et je vais dépenser cet hundred bucks de mon argent pour un spritz et une charcuterie chaque après-midi pour le reste de l'été.

Comment The vector art 3D rotate tool is super cool (Score 2) 7

I watched the "Project Turntable" video, and that is unbelievably cool.

To work, the AI has to recognize what the drawing represents, figure out how to reverse-transform the 2D representation into a 3D object, and work out what the hidden parts should look like. It's amazing.

It's only a short step between this and working out walk cycles, matching mouth movements to dialogue, adding facial expressions, etc.

This will revolutionize 2D animation.

Submission + - Samba gets funding from the German Sovereign Tech Fund.

Jeremy Allison - Sam writes: The Samba project has secured significant funding (€688,800.00) from the German
Sovereign Tech Fund (STF) to advance the project. The investment was
successfully applied for by SerNet. Over the next 18 months, Samba developers
from SerNet will tackle 17 key development subprojects aimed at enhancing
Samba’s security, scalability, and functionality.

The Sovereign Tech Fund is a German federal government funding program that
supports the development, improvement, and maintenance of open digital
infrastructure. Their goal is to sustainably strengthen the open source
ecosystem.

The project's focus is on areas like SMB3 Transparent Failover, SMB3 UNIX
extensions, SMB-Direct, Performance and modern security protocols such as SMB
over QUIC. These improvements are designed to ensure that Samba remains a
robust and secure solution for organizations that rely on a sovereign IT
infrastructure. Development work began as early as September the 1st and is
expected to be completed by the end of February 2026 for all sub-projects.

All development will be done in the open following the existing Samba
development process. First gitlab CI pipelines have already been running [4]
and gitlab MRs will appear soon!

https://samba.plus/blog/detail...

https://www.sovereigntechfund....

Comment they won't "fix" a proven marketing model (Score 1) 52

This is the exact same method by which Grubhub/Doordash/Postmates/etc get a large portion of their business: by getting their phone number and/or spoofed site listed as a search result or even in the google maps listing for a site nearest the searcher's location.
Of course Google won't "fix" the problem; it's a source of advertising revenue, and general boost to user engagement.

Comment Re:Maybe (Score 1) 104

The upstream Linux kernel doesn't differentiate between security bugs and "normal" bug fixes. So the new kernel.org CNA just assigns CVE's to all fixes. They don't score them.

Look at the numbers from the whitepaper:

"In March 2024 there were 270 new CVEs created for the stable Linux kernel. So far in April 2024 there are 342 new CVEs:"

Comment Re:Yeah (Score 1) 104

Yes ! That's exactly the point. Trying to curate and select patches for a "frozen" kernel fails due to the firehose of fixes going in upstream.

And in the kernel many of these could be security bugs. No one is doing evaluation on that, there are simply too many fixes in such a complex code base to check.

Comment Re:Maybe (Score 1) 104

You're missing something.

New bugs are discovered upstream, but the vendor kernel maintainers either aren't tracking, or are being discouraged from putting these back into the "frozen" kernel.

We even discovered one case where a RHEL maintainer fixed a bug upstream, but then neglected to apply it to the vulnerable vendor kernel. So it isn't like they didn't know about the bug. Maybe they just didn't check the vendor kernel was vulnerable.

I'm guessing management policy discouraged such things. It's easier to just ignore such bugs if customer haven't noticed.

Submission + - Why a 'frozen' distribution Linux kernel isn't the safest choice for security (zdnet.com) 1

Jeremy Allison - Sam writes: Cracks in the Ice: Why a 'frozen' distribution Linux kernel isn't the safest choice for security

https://ciq.com/blog/why-a-fro...

This is an executive summary of research that my colleagues Ronnie Sahlberg and Jonathan Maple did, published as a whitepaper with all the numeric details here:

https://ciq.com/whitepaper/ven...

Steven Vaughan-Nichols is covering the release of this
data here:

https://www.zdnet.com/article/...

Slashdot Top Deals

If the human brain were so simple that we could understand it, we would be so simple we couldn't.

Working...