This is an interesting choice; a big part of AI-mania has been corporations tripping over themselves to prove how “all-in” they are on AI. In firing a bunch of AI staff, Meta risks looking like they’re not committed enough to it. Are things starting to change? Is this all-in posture no longer important? (We can only hope.)
Nothing will meaningfully improve until the rich fear for their lives
Please pop the bubble already I’m so tired of the insanity
Aren’t we all?
This won’t happen but I would so love it if Nvdia was really brought to its knees and utterly dissolved. they so deserve it.
Nvidia is the only company making money off the AI blitz. This is a huge win for them, and they’re making money hand over fist. There is so little competition on the hardware side, and the software support for competitor’s products is weak in comparison, that they can charge basically whatever they want for their products.
The thing is that the frameworks for running things on competitors’ GPUs are actually fine (rocm and oneapi), the GPUs are price competitive or better as well. It’s just that CUDA/NVidia is the standard, and no one wants to learn a new language just to make something that most people aren’t going to be able to use. Very few people want to put in the effort to make something work across platforms.
There are some nice frameworks for general purpose GPU computing but it seems like they all have limitations of one type or another.
My hope is that the market crashes and there are SO MANY used chips out there that they can’t sell more
They just deserve to have CUDA published as a spec so others can compete.
Bad news. The AI bubble popping doesn’t mean AI goes away. If anything it will get worse from there.
Worse how?
There’s always an eviler fish.
Training AI is magnitudes more resource intensive than using AI. We don’t know what the final capabilities of GPT/etc will be when Open AI/etc gets liquidated to the highest bidders. Imagine if the guy who found Pandora’s box instead sold it at a blind auction before he opened it. Or something like that.
The global economy is undergoing rapid change on the level of post WW2 but without the unilateral guidance of the Marshall Plan. It’s not just responses to tariffs; initiatives in developing countries, increasing demand for rare earth minerals, and trends in energy production all make the future world economy hard to predict at this point.
AI slop is just what the public gets. We have to anticipate they aren’t telling us about the more insidious uses they’ve developed.
One can remain morally opposed to AI and skeptical of it’s ability to be authentically intelligent without sacrificing vigilance against it’s inevitable misuse.
I haven’t seen much evidence that it’s capable of much more than producing spam. I mean, that’s not harmless either, but yeah.
-
AI’s abilities aren’t static. The knowledge of the person using the AI is very much a factor. Case and point; I personally know a dev who bought a machine to run their own code agent and uses it to great success because they know how to break the task down and control quality.
-
It’s not how you prompt the task, it’s what tasks you prompt. AI is slop at human things… but computer things are easy mode. If used correctly it can allow 1 person to affect cyber attacks on a scale that would previously require entire teams of human hackers.
https://www.dhs.gov/sites/default/files/2024-10/24_0927_ia_aep-impact-ai-on-criminal-and-illicit-activities.pdf (Note: this is a pre-trump report)
My abilities aren’t static, either, but if I claimed that in a couple of years I’m going to be outrunning Sha’Carri Richardson, it would make sense for you to expect some evidence from me beyond just “I’m full of potential though!” before you really believed me.
If used correctly it can allow 1 person to affect cyber attacks on a scale that would previously require entire teams of human hackers.
Right, it’s doing spam. You can spam 1 gazillion people with phishing emails so much more effectively now. Like I said, that’s not harmless.
You can spam 1 gazillion people with phishing emails so much more effectively now.
-
Phishing is the most effect way to gain access to systems. So saying all it does is scale that up is not an argument against it’s threat whatsoever.
-
Phishing is only one of many types of attacks that AI can automate at scale… and arguably the easiest to detect.
-
The compounding factor is the ability to instantaneously cross analyze devices and software with a data base of known vulnerabilities. Do you know what a zero day exploit is?
-
This is what industry experts in the field are saying. Your analogy with running only holds water if you’re an authority on the subject and not a layman operating on conjecture.
-
-
chants Pop the bubble! Pop the bubble!
My god, US, have some self-respect and get some employment protection laws.
Just replace them with AI duh
They need the money for the data centers first – then the AI employees.
Energy is the real bottle neck. Data centers are a cup and ball game for Nvidia/OpenAI…
I think they could just do better and replace Zuckerberg with an LLM; He looks artificial; Often hallucinates, and flip flops on everything he says.
So it begins.
🫧💥
Careful Zucks, or they may question your commitment to Sparklemotion.
Is this for their ML team?
The only specific information I know is that this doesn’t effect their “super intelligence” team, and that they are inviting the laid off people to apply elsewhere at the company.
Ah yeah, instead of working to shift people elsewhere, lay them off and have them do the work and compete with each other as applicants again. Gotta love modern corporate America.
