nok22kon

8 ms·

91.
by nok22kon 4mo ago
programming is also solving problems in medicine they use all kinds of drugs which they don't really understand how they work. anesthetics is a great example
92.
by nok22kon 4mo ago
imagine installing an agent in slack at a company with 1000 employees, and you want each request to have its own VM for data analysis, downloading repos and working on them, ... regular VMs just use too much memory, a typical ubuntu uses 51
93.
by nok22kon 4mo ago
Capitalism and Artificial Intelligence are the same thing. As AIs grow stronger, they will be turned upon themselves - RSI. Nothing else matters. So everything is right on schedule, it was long predicted that the general public will never b
94.
by nok22kon 4mo ago
I think the Numogram is more interesting, highly relevant today due to AI happenings
95.
by nok22kon 4mo ago
parallel construction
96.
by nok22kon 4mo ago
we have a third category now - LLMs. this is where the value is captured now
97.
by nok22kon 4mo ago
Jobs come next after Tasks
98.
by nok22kon 4mo ago
the same argument was made 2 years ago: "in 2 years we'll be able to run GPT-4 level models on an expensive laptop, most people will be using this instead of the fancy cloud models". we are there, Gemma4/Qwen3.6 are GPT-
99.
by nok22kon 4mo ago
so you are against lending, stock markets and investing in general? because they are all based on "give me money today and I promise I will give you more money back tomorrow"
100.
by nok22kon 4mo ago
in Capitalism value=profit
101.
by nok22kon 4mo ago
obviously depends on your location and GPU for me it would be about $2 per day in electricity to generate 8 mil tok of Gemma4-26B at 4 bit quantization. this is excluding how much the GPU cost (no amortization) ignoring the fact that I coul
102.
by nok22kon 4mo ago
This is Capitalism working as intended - resources (RAM) are allocated to those which can extract most value from them (AI labs)
103.
by nok22kon 4mo ago
There is no subsidy on most OpenRouter providers, they are profitable today. You buy a big GPU, you serve LLMs, you print money.
104.
by nok22kon 4mo ago
those billion people which use LLMs every week, are they all being forced to use it?
105.
by nok22kon 4mo ago
https://www.theguardian.com/environment/2026/may/12/global-s...
106.
by nok22kon 4mo ago
but they move from tokenmaxxing to tokenmidding, not to tokenzero
107.
by nok22kon 4mo ago
RAM upgrade possibility has a downside though - very low RAM bandwidth, which is highly relevant today if you want to run local LLMs
108.
by nok22kon 4mo ago
better rule: exponentials are overlapping S-curves
109.
by nok22kon 4mo ago
if you do the electricity math you'll see that you pay more on local models while getting less (local is more heavily quantized) compared with OpenRouter. I'm not talking local Gemma/Qwen vs cloud Opus, but against OpenRouter
110.
by nok22kon 4mo ago
In most companies above 1000 employees it's assumed that any email/communication with most of the employees is as if you sent that communication to the NY Times at the same time. You are looking for a conspiracy where there is non
111.
by nok22kon 4mo ago
Sam Altman communicated it to the company. Somebody (or even the PR dept) leaked it to the press.
112.
by nok22kon 4mo ago
so I guess you would also support killing another TLD with a common filetype: https://ohmyz.sh
113.
by nok22kon 4mo ago
maybe you dont have GPU acceleration in your browser. that glass thing looks like its using complex blur effects
114.
by nok22kon 4mo ago
except they are extending it now for free
115.
by nok22kon 4mo ago
LLM are very good at looking at images and reasoning about them. much more than just object recognition/segmentation, they can explain the physics in the image, the intents, plan actions, ...
116.
by nok22kon 4mo ago
this is an LLM, not a motor cortex. it will output commands as text (json, ...), so comparing size is not very meaningful, especially considering neurons are highly complex and likely requires thousands of artificial simple neurons (weight+
117.
by nok22kon 4mo ago
that water is recirculated. nobody sends it down the drain after one loop
118.
by nok22kon 4mo ago
there is a glut if demand stops. this time demand doesn't stop. there is an exponential demand for tokens.
119.
by nok22kon 4mo ago
H100 were released in Oct 2022. They are now more expensive than at release time.
120.
by nok22kon 4mo ago
linux was plagued for a long time by lagging mouse cursor

More ›