Microsoft found a fitting way to punish AI for collaborating with SEO spammers in generating slop: make it use the GitHub code review tools. https://github.blog/changelog/2024-10-29-refine-and-validate-code-review-suggestions-with-copilot-workspace-public-preview/
Right, sure he picked the most glorified fantasy race out there, “but evil”. When in reality he’s a Duergar at best:
Personality Tyrannical, grim, industrious and pessimistic,[11][20] the lives of the gray dwarves were bleak and brutal. Rather than a flaw, however, they viewed their lack of happiness as their greatest strength, the defining feature of duergar pride.
Sounds about right for the joyless world he imagines his ideal society would represent.
The age of book shovelware has arrived: https://smartbitchestrashybooks.com/2024/10/ai-audiobook-narrators-in-overdrive-and-the-issue-of-library-ai-circulation-policy/
Edit: TIL about the US copyright office website’s ai generation info on works. That’s a thing we will be making a bunch of use of, I imagine.
My pal dug these up:
https://github.com/JusticeFighterDance/JusticeFighter110/issues/15 and
https://github.com/william-sto/JusticeNeverTooLate
They have issues providing commentary and caveats, and are in Chinese, ofc. And also take all of this with several boatloads of salt.
Apparently there are two GitHub repos up with strongly conflicting information on what happened, this is gonna be a three-buckets-of-popcorn incident.
They added sleeps to training jobs? Sounds like they deserve a raise for improving energy efficiency instead…
I would definitely hate to be burdened with an IQ. Such a terrible thing to be born with
A pal posed the theory that Matt Wordpress acts like he got ejected from some polycule and now I can’t unsee it
I’m temporarily-boating up to Boston
It’s the parable of the drowning man turning away all help because god will save him, all over again (except it’s floods and mudslides and forest fires)
Yes we do AGIle: we ask ChatGPT what our customers want and then ask it to write software to fulfill those requirements. Every two weeks it writes up some imaginary sprint retrospective meeting notes. Planning poker doesn’t work so well, this iteration isn’t so good at bluffing yet.
The industry called it “field engineering” previously, and “customer support” prior to that; renames happened every time the execs heard how this portion of their business is only a cost center and can easily be done by chat bots (to which the customer success people would say, good luck with that).
Eh, it’s destroying a common good to extract resources (landscape / minerals, electricity / numberwang). I say the shoe fits.
Cold fusion only needs another 10 years to bake, but then it’ll be awesome.
Hm, I don’t really see the sneer. They wrote a nasty bug, got notified and had a patch out for it within 36h. The remediations look reasonable too: better privacy, less firebase, actual security audits; even the bounty program is probably the right call (but they result in so many shit reports, it’s probably a wash).
I gotta admit I’m kind of partial to them and their browser? It’s the non-Brave one that ships with an Adblocker by default, has much nicer UI than the existing ones, and the sync thing isn’t half bad (if it doesn’t sync security badness to all your instances, ouch). Sure they sound like a cult but I guess that’s how browser dev gets funded since the 1990s.
I live like 15mi from there, I would prefer the containment bubble to stay intact. But the tech bubble is welcome to go blow up any moment
Let’s bring the haunted nuclear reactor back online so copilot can hallucinate a little more https://www.washingtonpost.com/business/2024/09/20/microsoft-three-mile-island-nuclear-constellation/
Hm, what do you mean by js-ware? That its front end uses JavaScript libraries? I guess, fair. Backend is python though (:
As a stunt (when I was unhappy with the previous linkding frontend), a pal and I wrote https://github.com/lz-bookmarks/lz, which is basically just linkding without the useful api and frontend (which is rust+webassembly, lol). Has a decent cli though, and interlinking between bookmarks and other URLs.
Despite efforts to remove overt racial prejudice, language models using artificial intelligence still show covert racism against speakers of African American English that is triggered by features of the dialect.
Got the pointer to this from Allison Parrish who says it better than I could:
> it's a very compelling paper, with a super clever methodology, and (i'm paraphrasing/extrapolating) shows that "alignment" strategies like RLHF only work to ensure that it never seems like a white person is saying something overtly racist, rather than addressing the actual prejudice baked into the model.
News 8's Meredith Jorgensen spoke with a victim and learned why a legal loophole may mean no one can be charged.
School student tells AI to put 20 other students’ faces on nude pictures, shares them in chat; it takes months for anyone including the school administrators to act because of some extremely, uh, dubious loophole.
If someone does that in photoshop, it’s a crime; if they do it in AI pretending to be photoshop, it’s somehow not. Gotta love this legal system’s focus on minor technicalities rather than the harm done.
They have Nik Suresh (the author) on, as well as Robert Evans. I haven’t listened to it all yet, but it’s fun so far.