864
The air begins to leak out of the overinflated AI bubble
(www.latimes.com)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 06 Sep 2024
864 points (97.4% liked)
Technology
59710 readers
2005 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
Thank fucking god.
I got sick of the overhyped tech bros pumping AI into everything with no understanding of it....
But then I got way more sick of everyone else thinking they're clowning on AI when in reality they're just demonstrating an equal sized misunderstanding of the technology in a snarky pessimistic format.
I’m more annoyed that Nvidia is looked at like some sort of brilliant strategist. It’s a GPU company that was lucky enough to be around when two new massive industries found an alternative use for graphics hardware.
They happened to be making pick axes in California right before some prospectors found gold.
And they don’t even really make pick axes, TSMC does. They just design them.
It's not trivial though. They also managed to lock dev with CUDA.
That being said I don't think they were "just" lucky, I think they built their luck through practices the DoJ is currently investigating for potential abuse of monopoly.
Yeah CUDA, made a lot of this possible.
Once crypto mining was too hard nvidia needed a market beyond image modeling and college machine learning experiments.
They didn't just "happen to be around". They created the entire ecosystem around machine learning while AMD just twiddled their thumbs. There is a reason why no one is buying AMD cards to run AI workloads.
One of the reasons being Nvidia forcing unethical vendor lock in through their licensing.
I feel like for a long time, CUDA was a laser looking for a problem.
It's just that the current (AI) problem might solve expensive employment issues.
It's just that C-Suite/managers are pointing that laser at the creatives instead of the jobs whose task it is to accumulate easily digestible facts and produce a set of instructions. You know, like C-Suites and middle/upper managers do.
And NVidia have pushed CUDA so hard.
AMD have ROCM, an open source cuda equivalent for amd.
But it's kinda like Linux Vs windows. NVidia CUDA is just so damn prevalent.
I guess it was first. Cuda has wider compatibility with Nvidia cards than rocm with AMD cards.
The only way AMD can win is to show a performance boost for a power reduction and cheaper hardware. So many people are entrenched in NVidia, the cost to switching to rocm/amd is a huge gamble
Imo we should give credit where credit is due and I agree, not a genius, still my pick is a 4080 for a new gaming computer.
Go ahead and design a better pickaxe than them, we'll wait...
Same argument:
"He didn't earn his wealth. He just won the lottery."
"If it's so easy, YOU go ahead and win the lottery then."
My fucking god.
"Buying a lottery ticket, and designing the best GPUs, totally the same thing, amiriteguys?"
In the sense that it's a matter of being in the right place at the right time, yes. Exactly the same thing. Opportunities aren't equal - they disproportionately effect those who happen to be positioned to take advantage of them. If I'm giving away a free car right now to whoever comes by, and you're not nearby, you're shit out of luck. If AI didn't HAPPEN to use massively multi-threaded computing, Nvidia would still be artificial scarcity-ing themselves to price gouging CoD players. The fact you don't see it for whatever reason doesn't make it wrong. NOBODY at Nvidia was there 5 years ago saying "Man, when this new technology hits we're going to be rolling in it." They stumbled into it by luck. They don't get credit for forseeing some future use case. They got lucky. That luck got them first mover advantage. Intel had that too. Look how well it's doing for them. Nvidia's position over AMD in this space can be due to any number of factors... production capacity, driver flexibility, faster functioning on a particular vector operation, power efficiency... hell, even the relationship between the CEO of THEIR company and OpenAI. Maybe they just had their salespeople call first. Their market dominance likely has absolutely NOTHING to do with their GPU's having better graphics performance, and to the extent they are, it's by chance - they did NOT predict generative AI, and their graphics cards just HAPPEN to be better situated for SOME reason.
This is the part that's flawed. They have actively targeted neural network applications with hardware and driver support since 2012.
Yes, they got lucky in that generative AI turned out to be massively popular, and required massively parallel computing capabilities, but luck is one part opportunity and one part preparedness. The reason they were able to capitalize is because they had the best graphics cards on the market and then specifically targeted AI applications.
His engineers built it, he didn't do anything there
As I job-hunt, every job listed over the past year has been "AI-drive [something]" and I'm really hoping that trend subsides.
"This is an mid level position requiring at least 7 years experience developing LLMs." -Every software engineer job out there.
That was cloud 7 years ago and blockchain 4
Reminds me of when I read about a programmer getting turned down for a job because they didn't have 5 years of experience with a language that they themselves had created 1 to 2 years prior.
Yeah, I'm a data engineer and I get that there's a lot of potential in analytics with AI, but you don't need to hire a data engineer with LLM experience for aggregating payroll data.
I'd argue there is a lot of potential in any domain with basic numeracy. In pretty much any business or institution somebody with a spreadsheet might help a lot. That doesn't necessarily require any Big Data or AI though.
The tech bros had to find an excuse to use all the GPUs they got for crypto after they bled that dry
If that’s the reason, I wouldn’t even be mad, that’s recycling right there.
I don't see a similar upgrade for "AI".
And I'm not a fan of BTC but $50,000+ doesn't seem very dry to me.
No, it's when people realized it's a scam