solenoid0937

6 ms·

61.
by solenoid0937 2mo ago
It depends what you mean by "early game." The first night cycle happens within minutes of starting and it's very interesting.
62.
by solenoid0937 2mo ago
> Why would we need to worry about China getting far ahead of US models If China gets to superintelligence first, the EU is cooked. This is a regime that has permanently disappeared tens of thousands of their human rights lawyers. > w
63.
by solenoid0937 2mo ago
This is a very easy question to answer: they're commoditizing AI and compressing margins for the US labs, while their state funds large scale buildouts. It's a decelerating force for the US because it removes dollars that the labs
64.
by solenoid0937 2mo ago
It is totally unrealistic to think that any country could regulate the other country's race to superintelligence. This is a technology that will give the victor control over the future of humanity forever, and the ability to lock out a
65.
by solenoid0937 2mo ago
You can use as many trendy "capitalism bad" phrases as you want, it doesn't change the reality that the EU has stagnated and fallen behind both economically and on the world stage, and will likely never catch up at this rate.
66.
by solenoid0937 2mo ago
This is a very naive viewpoint that does not understand that superintelligence is a race between the West and China. While the EU is twiddling its thumbs about copyright, it is shooting its own side in the foot during the most important rac
67.
by solenoid0937 2mo ago
Even in a hypothetical world where they somehow killed off US tech completely, they still wouldn't have their own actors develop. There is no collective will, no risk taking mindset. The VCs are cowardly with capital, and people don&#x
68.
by solenoid0937 2mo ago
I think you do not even have the faintest understanding of how good these models are getting and how quickly. It's not slowing down. That is exactly what they are building. You should make some friends at the labs to better inform your
69.
by solenoid0937 2mo ago
AI companies no longer talking about AI risk is a good thing now?
70.
by solenoid0937 2mo ago
They did not, you are confusing them with OpenAI which donated millions to Trump's inauguration fund.
71.
by solenoid0937 2mo ago
Just because you expect someone to do a bad thing doesn't mean you shouldn't shame them when they actually do the bad thing.
72.
by solenoid0937 2mo ago
It's a shame companies like Google immediately folded instead of standing with them. https://www.lesswrong.com/posts/iKm2FhpWkuuBojm82/why-i-left... In his recent interview Emil Michael (DoW undersecretary, &
73.
by solenoid0937 2mo ago
Do you have inside information we don't?
74.
by solenoid0937 2mo ago
If you can't prompt Claude to set up a symlink you probably shouldn't be using Claude.
75.
by solenoid0937 2mo ago
If it's overcomplicating and has poor explanations this seems like a harness/prompting issue? Fable is very steerable. For me it makes the "simplest thing that works" and - once I tuned it - very understandable explanati
76.
by solenoid0937 2mo ago
IDK I use open models every day for personal projects, and closed models for work. Open models are all decidedly far behind Fable and a good bit behind Opus as well. All of these posts read like motivated/wishful thinking to me. I get
77.
by solenoid0937 2mo ago
Reddit is an incredibly easy website to bot/psyop, especially if you have older accounts.
78.
by solenoid0937 2mo ago
Their margins are already quite high and this is a play to bleed customers from Anthropic, they will of course raise prices after IPO.
79.
by solenoid0937 2mo ago
Unless it's in your contract, they will use the data. They might not be using it now , but they will eventually.
80.
by solenoid0937 2mo ago
There is no Claude hype on HN, HN has more vitriol for Claude than any online forum I've seen
81.
by solenoid0937 2mo ago
Don't all these AI companies run at negative margins? How do they afford to buy all the books in the world? I don't see how these companies are not already bankrupt given how much they subsidize
82.
by solenoid0937 2mo ago
Sure, but front ends like LMStudio exist for that crowd Otherwise, if you're a programmer setting up a local harness, it only takes like 20-30 minutes to learn what the right parameters are. It's very model, hardware, and use case
83.
by solenoid0937 2mo ago
Sure but that's true with the closed models as well. They run on different providers and you can choose the cloud provider you prefer.
84.
by solenoid0937 2mo ago
Open source software does not require trillions of dollars in R&D to produce the next iteration
85.
by solenoid0937 2mo ago
Literally nothing in their proposal is about blocking access to Chinese models, HN has devolved so much with these conspiratorial hot takes.
86.
by solenoid0937 2mo ago
> we cannot step debug an llm's output to find out what happened We absolutely can with mechanistic interpretability & companies like Anthropic, OpenAI, Meta, and Google do precisely this do debug their models.
87.
by solenoid0937 2mo ago
I know world-leading experts in their field that avoid oral debates precisely because they don't think on their feet.
88.
by solenoid0937 2mo ago
This is a nightmare for students that don't think on their feet.
89.
by solenoid0937 2mo ago
No, its that going from $2k/mo to $20/mo is not worth the capability loss in the slightest, because AI just isn't a big part of our fixed costs.
90.
by solenoid0937 2mo ago
Expecting profit during hypergrowth is silly

More ›