Rendered at 23:31:32 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
iamflimflam1 7 hours ago [-]
Man whose job depends on you not running local models doesn’t want you running local models…
tompark 6 hours ago [-]
> “A lot of people want local inference,” Altman said. “And somehow, they assume that the problem is solved now. Because we are going to have open-source models. We’re going to be able to run them on our own computers. And so we’re not going to need all the stuff [DRAM] in the data center.”
At first TFA seemed like typical tech news but reading this quote suddenly struck me as satire, as if it came from The Onion.
Said with old-timey incredulity: "I quite enjoy the bread line, an oven at home? Could you imagine the risk/waste?"
jwilcoxson 7 hours ago [-]
That guy sucks.
mindslight 6 hours ago [-]
After a year of finishing projects with storage bought before the start of the second grump catastrophe (and seeing that everything has basically gone up in price 2.5x), I just had to break down and buy some DDR5 SODIMMS (which I had unfortunately not stocked up on). It hurt, and I am definitely wondering if this is going to look foolish in retrospect - ie buying at the peak.
But then I remembered that it's essentially a hedge. These people and the capital pools behind them give off strong vibes of being all-in on capturing computing technology for their centralized control, and damning the revolutionary distributed computation/communications we have been enjoying for the past three decades (albeit we've been squandering most of its capabilities).
downrightmike 7 hours ago [-]
That's why we'll get laptops with hbm installed
RetroTechie 6 hours ago [-]
His logic is flawed in multiple ways:
1) Assuming that RAM in AI datacenter is doing more useful work than outside it.
Wrong! That depends entirely on what it's used for. And since mr. Altman doesn't KNOW what your or my RAM chip(s) are used for, he also can't decide which use is more important.
2) What he deems more important, is a matter of opinion. What's important to him, likely won't matter to me. And vice versa.
Go read some history books, mr. Altman! Maaany wars have been fought over "I know what's best for everyone! Now bow & kiss my toes!"
Oh and 3) all those AI datacenters are built using other people's money than the tech bros running the show. Whereas consumers pay for RAM with their own money. But hey, no doubt it's more fun to shovel *other* people's cash into a firepit.
At first TFA seemed like typical tech news but reading this quote suddenly struck me as satire, as if it came from The Onion.
But then I remembered that it's essentially a hedge. These people and the capital pools behind them give off strong vibes of being all-in on capturing computing technology for their centralized control, and damning the revolutionary distributed computation/communications we have been enjoying for the past three decades (albeit we've been squandering most of its capabilities).
1) Assuming that RAM in AI datacenter is doing more useful work than outside it.
Wrong! That depends entirely on what it's used for. And since mr. Altman doesn't KNOW what your or my RAM chip(s) are used for, he also can't decide which use is more important.
2) What he deems more important, is a matter of opinion. What's important to him, likely won't matter to me. And vice versa.
Go read some history books, mr. Altman! Maaany wars have been fought over "I know what's best for everyone! Now bow & kiss my toes!"
Oh and 3) all those AI datacenters are built using other people's money than the tech bros running the show. Whereas consumers pay for RAM with their own money. But hey, no doubt it's more fun to shovel *other* people's cash into a firepit.