Lemnear wrote: ↑Sat Feb 01, 2025 12:23 pmwell then it means that Sam Altman "defrauded" investors...and not a little.
....sigh.
....
sigh.
Yes Altman is a sales guy. Yes he wears $600,000 wrist-watches like Ellen DeGeneres. Yes he's a vampire ghoul like Steve Jobs or Bill Gates. Yes his goal is grab as many resources and money as possible. A sales guy who cares about anything other than money is a bad sales guy. Everyone knows that: If a man won't sell his grandma for a buck fiddy he has no business being a sales guy.
In Hinton's interview when he won the nobel prize for physics, he said he was most proud of his student, Ilya, firing Altman.
Something all these people who don't believe in/don't understand AI or have their own grift have been saying for decades is neural nets 'would never be useful for anything'. And the rebuttal has always been "Yes, because the computer hardware isn't good enough to make anything useful for humans yet. It's 1960."
The only reason they're able to do anything filthy normies care about now is scale. The state of the art in image generation a decade ago was
StackGAN. Followed up a couple years later by StyleGAN. Yeah, a few more approaches to things have been developed. But the core difference is
scale. If you don't believe in scale, you don't believe in AI. Or that large brains in animals are necessary to have more capabilities.
One of the biggest red flags that a person talking is has no business in being in the discourse is how often they pretend text predictors are the only thing AI can and ever will be able to do (what the fuck ever happened to hand-writing recognition?! That and, then voice-to-text were always the big things in the old days! See what I mean about trend-chasers, they're not even familiar with the most basic ancient history! They only have enough room in their brains for CURRENT_THING. OLD_THING and UPCOMING_THING are the stuff of myths.). If you see the sentence "You can't get AGI by making GPT-7!", just ignore the clown and walk away.
The goal isn't to fit a single curve. You can't make an AGI with a single curve, everyone knows that. The goal is to fit all the curves. To do that, you need scale large enough to hold them all.
As for the poor wittle investors, most of them are indeed screwing themselves, but not because the god computer constantly pumping out a million years worth of technological advancement is going to be rendered obsolete by someone's souped-up raspberry pi running a one token a second output on a text model that was state of the art in 2021. It's because money is a control mechanism for human labor, and the ultimate end goal here is to replace that. Their website even has a big fuck-off warning in a giant magenta box warning people about this:
This is obviously not something Mr.Sam brings up unprompted. (But some might take delight in the seed money they scammed Musk out of as a 'donation'. They were wily enough to NOT let him be the CEO and ruin everything, unlike certain other companies. : D)
Most of them have no idea what they're investing in or what it really means, they're just blindly chasing trends and their own greed. One of the most basic, obvious questions that always pops up: "If nobody has a job, how you gonna make money?" There are serious responses to that, like how Wal-Mart is heavily funded by welfare, and like over 30% of our economy already is driven by welfare... but by and large, it isn't about money. Money is the feedlot for us little cattle, it's not a means to an end in itself. It's about having power and things.
Robot army > Not having a robot army. It's as simple as that. If it's physically possible, it will be done. It isn't going to 'magically come out of nowhere'; we have hundreds of billions of annual investment and a shitton of people working to make it happen. AGI will be built in a giant fuck-off datacenter, and some years later the infrastructure will be complete to pump out NPU's for the first generation of true sci-fi robots.
Once again,
the state of the art in robotics nine years ago.
The difference between now and ten years from now will be wider than it was between now and ten years ago. Anybody who has the slightest clue of what the numbers involved are knows that.