The problem is that there’s no incentive for employees to stay beyond a few years. Why spend months or years training someone if they leave after the second year?
But then you have to question why employees aren’t loyal any longer, and that’s because pensions and benefits have eroded, and your pay doesn’t keep up as you stay longer at a company. Why stay at a company for 20, 30, or 40 years when you can come out way ahead financially by hopping jobs every 2-4 years?
Ah, yes, you don’t have an actual rebuttal so everything is just “propaganda” and “cyberpunk dystopia” as if snake oil salesmen hawking freaking AI-powered vibrators and vagueposting about the benefits of AI while downplaying or ignoring its very real, very measurable harms, while an entire cottage industry of individuals making a living on their creative endeavors being forced into wage slave office jobs isn’t even more of a dystopia.
Try actually talking to an artist sometime bud, I don’t know of a single one that is actually okay with AI, and if you weren’t either blind or an “ideas guy” salivating at the thought of having a personal slave to make (shitty, barely functional, vapid) shit without paying someone with the actual necessary skills, you’d agree too.
ideally? It means that AI companies have to throw away their entire training model, pay for a license that they may not be able to afford, and go out of business as a result, at which point everyone snaps out of the cult of AI and realizes it’s as overhyped as block chain and pretends it never happened. Pardon me while I find a flea to play the world’s tiniest violin. More realistically, open models will be restricted to FOSS works and the public domain, while commercial models pay for licenses from copyright holders.
Like, what, you think I haven’t thought through this exact issue before and reached the exact conclusion your leading questions are so transparently pushing that open models will be restricted to public works only while commercial models can obtain a license? Yeah, duh. And you know what? I. Don’t. Care. Commercial models can be (somewhat) more easily regulated, and even in the absolute worst case, at least creators will have a mechanism to opt out of the artist crushing machine.
Yeah, no, stop with the goddamn tone policing. I have zero interest in vagueposting and high-horse riding.
As for what I want, I want generative AI banned entirely, or at minimum restricted to training on works that are either in the public domain, or that the person creating the training model received explicit, opt-in consent to use. This is the supposed gold standard everyone demands when it comes to the widescale collection and processing of personal data that they generate just through their normal, everyday activities, why should it be different for the widescale collection and processing of the stuff we actually put our effort into creating?
Huh? How does that follow at all? Judging that the specific use of training LLMs–which absolutely flunks the “amount and substantiality of the portion taken” (since it’s taking the whole damn work) and “the effect on the market” (fucking DUH) tests–isn’t fair use in no way impacts parody or R34. It’s the same kind of logic the GOP uses when they say “if the IRS cracks down on billionaires evading taxes then Blue Collar Joe is going to get audited!”
Fuck outta here with that insane clown logic.
It’s like nobody here actually knows someone who is actually creative or has bothered making anything creative themselves
I don’t even have a financial interest in it because there’s no way my job could be automated, and I don’t have any chance of making any kind of money off my trash. I still wouldn’t let LLMs train with my work, and I have a feeling that the vast majority of people would do the same
So because corps abuse copyright, that means I should be fine with AI companies taking whatever I write–all the journal entries, short stories, blog posts, tweets, comments, etc.–and putting it through their model without being asked, and with no ability to opt out? My artist friends should be fine with their art galleries being used to train the AI models that are actively being used to deprive them of their livelihood without any ability to say “I don’t want the fruits of my labor to be used in this way?”
Hell, that article is also all about Google Books, which is an entirely different beast from generative AI. One of the key points from the circuit judge was that Google Books’ use of copyrighted material “…[maintains] respectful consideration for the rights of authors and other creative individuals, and without adversely impacting the rights of copyright holders.” The appeals court, in upholding the ruling that Google Books’ use of copyrighted content is fair use, ruled “the revelations do not provide a significant market substitute for the protected aspects of the originals.”
If you think that gen AI doesn’t provide a significant market substitute for the artwork created by the artists and authors used to train these models, or that it doesn’t adversely impact their rights, then you’re utterly delusional.
An actual technical answer: Apparently, it’s because while the PS5 and Xbox Series X are technically regular x86-64 architecture, they have a design that allows the GPU and CPU to share a single pool of memory with no loss in performance. This makes it easy to allocate a shit load of RAM for the GPU to store textures very quickly, but it also means that as the games industry shifts from developing for the PS4/Xbox One X first (both of which have separate pools of memory for CPU & GPU) to the PS5/XSX first, VRAM requirements are spiking up because it’s a lot easier to port to PC if you just keep the assumption that the GPU can handle storing 10-15 GB of texture data at once instead of needing to refactor your code to reduce VRAM usage.
More like they were a darling up until they were compromised by Russian intelligence and turned into the propaganda arm of the protofascist party in the US.
It makes sense to judge how closely LLMs mimic human learning when people are using it as a defense to AI companies scraping copyrighted content, and making the claim that banning AI scraping is as nonsensical as banning human learning.
But when it’s pointed out that LLMs don’t learn very similarly to humans, and require scraping far more material than a human does, suddenly AIs shouldn’t be judged by human standards? I don’t know if it’s intentional on your part, but that’s a pretty classic example of a motte-and-bailey fallacy. You can’t have it both ways.
Still a copyright violation, especially if you make it publicly available and claim the work as your own for commercial purposes. At the very minimum, tracing without fully attributing the original work is considered to be in poor enough taste that most art sites will permaban you for doing it, no questions asked.
Smaller doesn’t matter if they’re going in a 3.5" tray. There are some models that only come with 2.5" trays, but go figure, the only 2.5" model that isn’t a 5-figure all-flash enterprise-scale model is one of our least popular models
I work tech support for a NAS company and the ratio of HDDs to SSDs is roughly 85-15. Sometimes people use SSDs for stuff that requires low latency, but most commonly they’re used as a cache for HDDs in my experience.
You need to learn your Internet history. It wasn’t so long ago that we had a diverse, interoperable community of instant messaging platforms based on XMPP, an open, federated protocol. Anybody could host their own XMPP server, and communicate with any other XMPP server. Then in 2006, Google added XMPP support to their Talk app and integrated it into the Gmail web interface. But there were problems:
First of all, despites collaborating to develop the XMPP standard, Google was doing its own closed implementation that nobody could review. It turns out they were not always respecting the protocol they were developing. They were not implementing everything. This forced XMPP development to be slowed down, to adapt. Nice new features were not implemented or not used in XMPP clients because they were not compatible with Google Talk (avatars took an awful long time to come to XMPP). Federation was sometimes broken: for hours or days, there would not be communications possible between Google and regular XMPP servers. The XMPP community became watchers and debuggers of Google’s servers, posting irregularities and downtime (I did it several times, which is probably what prompted the job offer).
And because there were far more Google talk users than “true XMPP” users, there was little room for “not caring about Google talk users”. Newcomers discovering XMPP and not being Google talk users themselves had very frustrating experience because most of their contact were Google Talk users. They thought they could communicate easily with them but it was basically a degraded version of what they had while using Google talk itself. A typical XMPP roster was mainly composed of Google Talk users with a few geeks.
Only a few years later, Google would discontinue Google Talk, migrated all their users to Hangouts, and decimated the XMPP community in an instant. Most of the Google users never noticed, outside of some invalid contacts in their list.
That’s why everyone distrusts Meta. Even with Threads being a relatively unsuccessful platform by commercial social media standards, its active userbase still dwarfs the entire Fediverse combined. There’s absolutely nothing stopping Meta from running the exact same playbook:
Add ActivityPub support, but only partially
Add new features to ActivityPub without consulting with the rest of the Fediverse or documenting the extensions, degrading the experience for everyone not using Threads
Entice Fediverse users to migrate to Threads–after all, why use Mastodon or Lemmy when 95%+ of ActivityPub traffic originates from Threads?
Deprecate ActivityPub support after most of the Fediverse is on Threads, leaving it smaller and more fragmented than if Threads had never federated at all, while forcing everyone who migrated from another Fediverse platform to Threads into an impossible choice between abandoning the vast majority of their contacts or subjecting themselves to Meta’s policies, tracking, and moderation
You think there are curves on this abomination?
Who even knows? For whatever reason the board decided to keep quiet, didn’t elaborate on its reasoning, let Altman and his allies control the narrative, and rolled over when the employees inevitably revolted. All we have is speculation and unnamed “sources close to the matter,” which you may or may not find credible.
Even if the actual reasoning was absolutely justified–and knowing how much of a techbro Altman is (especially with his insanely creepy project to combine cryptocurrency with retina scans), I absolutely believe the speculation that the board felt Altman wasn’t trustworthy–they didn’t bother to actually tell anyone that reasoning, and clearly felt they could just weather the firestorm up until they realized it was too late and they’d already shot themselves in the foot.
…So your metric of “too much AI safety” is that it won’t let you fuck the fish…?
The speculation I heard in the Ars Technica article is that the board was unhappy with how quickly he was pushing to commercialize OpenAI, and they were wary about all the AI side hustles he was starting, including an AI chip company to compete with nvidia.
Why make a better UI when it’ll probably introduce a slew of new bugs?