fwipsy

6 ms·

61.
by fwipsy 2mo ago
Each spark has as much VRAM as the server built in the article.
62.
by fwipsy 2mo ago
Are you sure that AI hasn't just saturated your personal benchmark? Maybe you're just not asking it to do things which showcase its full capability. Open-source models will catch up for any given use case, but the frontier is inte
63.
by fwipsy 2mo ago
To me it indicates that he knew how easy it is to lose your shirt, likely from the examples of his peers -- and then he did anyways.
64.
by fwipsy 2mo ago
> In recent times, Johannes Heidecke, Head of OpenAI's Safety Systems team, also left the company – as did OpenAI’s Chief Futurist Joshua Achiam. It's OK though, they don't need an ethics team because everyone at OpenAI is
65.
by fwipsy 2mo ago
Why do you think cloud AI doesn't also use these optimizations? They are also compute-constrained. I expect optimization and scale have a multiplicative effect. Surely all the serious players care about both.
66.
by fwipsy 2mo ago
The robot market would have to be pretty damn valuable, to be worth throwing away a lead in the model market. Say, 10-100x as valuable.
67.
by fwipsy 2mo ago
It's pretty ironic to read about Twain losing his fortune to bad investments, after satirizing sketchy investments in "The Gilded Age."
68.
by fwipsy 2mo ago
14mb? More like sub-$20 devices.
69.
by fwipsy 2mo ago
Maybe you're right for smaller models, but for frontier models this is the tail wagging the dog. If China has exclusive frontier model capability in the future, that's an enormous commercial and geopolitical lever. They won't
70.
by fwipsy 2mo ago
Good points, I personally believe that if/when China takes the lead, they will immediately stop releasing model weights. It only makes sense as a strategy to counterbalance (current) American labs' monopoly on frontier models. Hol
71.
by fwipsy 2mo ago
Part of negotiating is understanding when no compromise is possible. Cancelling the rental seems like the best outcome here, although I don't like the author's tone much.
72.
by fwipsy 2mo ago
FWIW I've also had success getting AI to write readable code. I guess my point is that allowing it to write opaque code may leave room for unintended behavior.
73.
by fwipsy 2mo ago
I took it for granted that as AIs got smarter, their taste in code would get better. The alternative seems much worse. If AIs get smarter but the code remains opaque, then eventually there may be no way for humans to verify it. This seems e
74.
by fwipsy 2mo ago
"We're the good guys because we'll destroy the world a little less."
75.
by fwipsy 2mo ago
I think Anthropic employees think that their product is more like opiates -- highly dangerous, but with a large potential benefit when applied correctly. I don't know what OpenAI employees were thinking, but thankfully it looks like th
76.
by fwipsy 2mo ago
The culture at frontier labs is set by people who have been in the field for over a decade--AI's true believers, who expect it to be a technology as dangerous and disruptive as nuclear weapons. They build it anyways because they think
77.
by fwipsy 2mo ago
Well ackshually... > Beaver food flavoring doesn't actually come from the animal's anal gland but from another organ next to it, called the castor sac. I still wouldn't try it, myself.
78.
by fwipsy 2mo ago
CYA. This approach not working is "the user's fault."
79.
by fwipsy 3mo ago
> pitch depends on position of hinge > volume depends on how fast you adjust it Yeah, I had the same thought - in all the creaky doors I've seen, pitch and speed go together.
80.
by fwipsy 3mo ago
Neat! There's a notable omission though--none of the earth globes look like earth does from space, with true color* AND clouds. A while ago I got really into making paper globes because I wanted one like that and I couldn't find o
81.
by fwipsy 3mo ago
I thought of the same series, but that's not the tl;dr I would have chosen at all. I would have gone to part 4e where I found this quote near the end: > We should also note that these women work an enormous amount. We estimated, you
82.
by fwipsy 3mo ago
Hey, it's all good. Sometimes things get a little heated, but I still feel like you were basically engaging in good faith. I should have skipped straight to the point and found the source in my first reply. Also, I probably overstated
83.
by fwipsy 3mo ago
GLM 5.2 didn't defend them at all. It only helped with the postmortem. HF was fucked either way. The only thing that will really prevent this is tighter restrictions on the attacking model. We need policies which asymmetrically help de
84.
by fwipsy 3mo ago
I apologize; it was lazy of me not to find a source. This one specifically says that the model was Fable; does not mention which model they attempted to access on the OpenAI side: https://www.cnbc.com/2026/07/24&#x
85.
by fwipsy 3mo ago
Amodei isn't ignoring reality; he's just proposing a different solution to the problem. If it were one of Anthropic's models, then that would be a much stronger case. Rapid proliferation of hacking capabilities may make exper
86.
by fwipsy 3mo ago
I assume you mean https://huggingface.co/blog/security-incident-july-2026 . It says that they used frontier models, not frontier cybersecurity models. My reading is that they asked Fable and it refused; if they'd h
87.
by fwipsy 3mo ago
> an advanced, closed, American AI model hacked another company. The only defense was open-source AI from China. HuggingFace did not seek access to Claude Mythos or OpenAI's equivalent program. They probably could have had access to
88.
by fwipsy 3mo ago
In some places in Europe they're on the far left. I know Trump has a bit of a grudge against Europe, maybe that's what this is about.
89.
by fwipsy 3mo ago
The same perverse incentives apply to the stock market too. "I'm short United Healthcare, someone should go shoot their CEO." Or "I'm long defense stocks, I guess I'll vote for someone who will start a war.&quo
90.
by fwipsy 3mo ago
Knowing whether a major pandemic has a 2% chance vs. 0.2% chance per year has major policy implications. Just like the difference between a 100-year floodplain and a 500-year floodplain has important insurance implications, even though you

More ›