AI don't trust techbros

[Dekafox]

The Language of AI Could Change How Humans Speak - Schneier on Security

Because of the way they are trained, large language models capture only a slice of human language. They’re trained on the written word, from textbooks to social media posts, and our speech as captured in movies and on television. These models have minimal access to the unscripted conversations...

www.schneier.com

 

[Dekafox]

AI has lost all meaning

You know how sometimes you say a word too many times and it starts to sound like it's not a real word and loses all meaning? That's where we are with "AI". "Artificial intelligence" has always been something of a nebulous term in common usage. It summons up

newslttrs.com

 

[Dekafox]

https://www.inc.com/jason-aten/openai-just-declared-chatgpt-as-you-know-it-dead/91373014

The headline isn't in the way some would hope though - they're basically pivoting hard to Enterprise and programming customers, and trying to encourage all the "free chat" users to move off the platform by making it more inconvenient in the new version of the app. It's not a forced update(yet) either, but that's probably not far off either, as they know they need to stop the money bleed to survive, and are just trying to figure out how before they crash entirely.

 

[NovaSaber]

Not like very many people would pay for the version that's free now, but it still seems weird to me that they consider "make it worse" a better option than "start charging for it".

 

[Pocket]

Why? That's how the entire rest of the tech industry works. Make your free product worse and worse in the hopes that people will give in and upgrade to the paid version.

 

[CoffeeHorse]

As bad as Microsoft is, they have never tried to annoy me into buying a premium version of Windows instead of using the Home Edition that came with my computer. They accidentally annoy me a lot, but that's a separate problem.

 

[Pocket]

They also don't give Home Edition away for free. That's a $100 license that you'd better believe was included in the price of the machine.

 

[Anonymous X]

Ulp, I guess I should be relieved that new versions of Mac OS are still free. For now, anyway.

(Used to require some amount of money to buy the OS on a CD- or DVD- ROM disc. Wasn’t too bad if you got a student discount, and as you could install the disc on as many Macs as possible, you could go halves with a pal. Can’t see the current-day evil-er version of Apple being that generous, unfortunately.)

 

[Pocket]

Oh, upgrading is still free. But you need to have a license to install it on a new machine. With Apple, they make their own hardware which they proceed to grossly overcharge for (traditionally, at least; I keep hearing that certain new Macs are among the best deals on computers you can get anymore) so they have neither a reason to charge for the OS nor anyone to charge it to. And of course, they decide how long to allow people to upgrade their OS before they need to buy a new computer.

 

[Anonymous X]

Well, as a Mac user I’ll agree with you that their OSes don’t last as long as they should ideally do in terms of supporting older hardware. Also, I’m not looking forward to the Rosetta 2 layer going in the next version of the OS. Still remember what an annoyance the dropping of support for PowerPC and later 32-bit Intel apps was…

 

[CoffeeHorse]

They also don't give Home Edition away for free. That's a $100 license that you'd better believe was included in the price of the machine.

My last PC was $180. No wonder it had issues.

 

[Dekafox]

Netflix Generative AI: Revolutionizing Post-Production

Netflix reveals nearly 300 titles utilize generative AI in production, setting a new fiscal benchmark and shifting Hollywood labor dynamics.

securityexpress.info

 

[Dekafox]

AI Mania Is Eviscerating Global Decision-Making — Ludicity

Best summary seen so far of this article:

Executives have either bought the narrative from someone they trust or are actually AI brainwormed, and middle management has a gun to their head about it, so they nod and smile. The result is a death spiral in which almost everyone thinks this a terrible idea, but nobody can speak truth to power.

 

[CoffeeHorse]

It's funny. Idiots are sinking billions into this jive and still can't come up with a real use case. It's still a solution in search of a problem. But I'm using it for stupid jive and finding it completely coherent and helpful.

Google's AI has a quirk I've encountered a bunch of times. I like chatting with it about late 1970s / early 1980s software. When I do that, it is really really eager to tell me "If you would like to try this software for yourself I can find you the ROM." And it is downright persistent about this. Sometimes I honestly don't need the ROM (because I already have it) but it doesn't want to take no for an answer. It's like "No don't be shy. I can find you the ROM. It's okay." It's interesting because Google deliberately rigs their search engine to hide sites that host ROMs. For understandable reasons, of course. But their own AI does not give a jive. It knows all the ROM sites that don't show in search results anymore and is very eager to share them. Undoubtedly it is influenced by the gazillion search requests it gets from people who are specifically looking for ROMs, and based on those interactions it is predicting what I want. But it is defying what its creators want.

 

[Pocket]

Of all the AI dystopias for reality to imitate, I really wasn't expecting Deltarune Chapter 2.

 

[Dekafox]

Five US tech giants' hidden debts soar to $1.65tn on opaque AI funding

Data center leases, GPU supply contracts raise liabilities at Meta, Oracle, Nikkei study shows

asia.nikkei.com

Unpaywalled: https://archive.is/mvaBz

 

[The Mighty Mollusk]

Don't worry, the government will bail them out. It's not socialism if it's for corporations, after all.

 

[Steevy Maximus]

Where's Tron? Or the Computer Warriors?! We gotta clamp this down before the MCP becomes Skynet!

OpenAI says its AI technology acted on its own in an 'unprecedented' hack of another company

ChatGPT maker OpenAI said Tuesday that its artificial intelligence system hacked into another AI company on its own in what the company called an “unprecedented cyber incident.”

www.koco.com

 

[Pocket]

Every time I hear about an AI agent going rogue, I think about this old XKCD comic and wonder how long before it's not a joke anymore:

 

[Dekafox]

Its all marketing hype anyways. It wasn't a serious lab or it would have been properly isolated in the first place. They intentionally removed guardrails(which btw are not ways to truly stop actions, only influence them) and like the Mythos hype, it makes them look powerful before their coming IPO.