vb-8448

8 ms·

181.
by vb-8448 5mo ago
I bet that in 10 years they will name the next financial (and perhaps economic) crisis "the subsidized bubble" Until they find someone else paying the costs it will not pop.
182.
by vb-8448 5mo ago
Is this a rhetorical question? The answer seem so obvious. It is also obvious that people will try to understand how to hack the whole algorithm ... we are entering in a new era of hot air SEO like experts.
183.
by vb-8448 5mo ago
You are confusing "slop" with "information", there is so much slop because it costs nearly 0 to be produced, but there's far less "information" than you are thinking.
184.
by vb-8448 5mo ago
> We built it, because we as humans intrinsically know that information should be free I don't know if this statement is more stupid or naive ..
185.
by vb-8448 5mo ago
I guess it's most appropriate so say "LOSSY COMPRESS".
186.
by vb-8448 5mo ago
Maybe there are some exceptions, but my[0] behaviour changed a lot in the last 3 years: - in the past Google was the entry point for everything, I was opening every single site at the top of my search and navigate through it. E.g.: if the s
187.
by vb-8448 5mo ago
Actually it cannot work this way, content creators make far more money from ads in the video itself compared to the one yt gives them. If it were for yt money alone basically we will still be in the 2010 yt: folks that doing it just for fun
188.
by vb-8448 5mo ago
The fact is that internet is already "tech giants own realm": the power is way beyond public imagination and affects all of us in real life on daily bases, but there are still people thinking they are not the "evil one"
189.
by vb-8448 5mo ago
I, and I guess basically everyone here, don't have access to OAI or Anthropic books, and it's really difficult to disprove your statements but: - AI revenue going up & cost/token are not related metrics, at least not in t
190.
by vb-8448 5mo ago
Not consumer hw ... and if we speak about local llm we cannot assume most of us can put a rack in their basement Already today is not possible to run deep seek v4 pro locally, and I cannot imagine that in 2 years we will be.
191.
by vb-8448 5mo ago
Today open weights frontier models cannot run locally, unless quantization is used. Deep seek v4 pro require almost 1 TB of RAM in INT4. I hardly doubt there will be consumer grade HW to run it in 2 years either. And deep seek v4 pro is not
192.
by vb-8448 5mo ago
Is sounds very much like "trust me bro" ... Obviously I, like basically everyone else here, don't have access to Open AI or Anthropic books so it's just guessing based on public available evidences, but "tokens aren
193.
by vb-8448 5mo ago
do you have some ref?
194.
by vb-8448 5mo ago
> within a few years we will be running local models as good as today’s frontier Unless there isn't some important breakthrough in hw production or in models architecture, it's quite the opposite: bigger, more expensive and mor
195.
by vb-8448 5mo ago
His point looks still correct (the total revenue reported to the public is 34% higher than the one in the affidavit) but the title is definitely clickbait. I guess the "5b vs 6.67b" pushes view counters less than "5b vs 19b&q
196.
by vb-8448 5mo ago
Here[0] the original affidavit, IANAL but I'm pretty sure "up to date" in this context means "8th of March". [0] https://storage.courtlistener.com/recap/gov.uscourts.cand.46...
197.
by vb-8448 5mo ago
To be honest I'm pretty sure he is right about "AI players are basically throwing out numbers because the real one are not so shiny (yet?)". What I'm not really fan about is the communication style and the "trust me
198.
by vb-8448 5mo ago
> they might report the number in the court affadavit with one month lag for various reasons According to the article the total revenue figure in the affidavit is “to date” , so basically 8th or March.
199.
by vb-8448 5mo ago
The $19b ARR is from March 3rd, so it is based on the revenue from February.
200.
by vb-8448 5mo ago
"Exceeds $5b" is a weird way to say "5$b +34%" ... Don't get me wrong, I'm not a big fan of Zitron, but the narrative around revenues and profitability around AI labs are quite misleading.
201.
by vb-8448 5mo ago
I guess at some point we will have lawyers, attorneys and judges using this stuff ... at the point lawyers will become kinda "seo"/"copywriter" experts on how to better trick the others LLM.
202.
by vb-8448 5mo ago
In general: less data = less "intelligence". And basically all the security bugs I've read about were find looking on the source code. But it doesn't mean windows is more secure, just image a scenario where someone is st
203.
by vb-8448 5mo ago
I guess we miss fundamental information: how much in terms of time and token usage took to the "middle guy" to create the report? Next question: could it be that OP can use Mythos in a better way since he knows better the project?
204.
by vb-8448 5mo ago
With SOTA models I can just set up the instructions (even a little bit fuzzy), go away for 10 or 15 minutes, come back and just check result and adjust when necessary (and most of the time small adjustment are necessary, but the overall wor
205.
by vb-8448 5mo ago
Actually even with a 9k hardware you won't get good enough performance. There is an interesting video from antirez on trying to run deepseek v4 flash 2bits on a m3 max 128GB ... and the result is kind delusional: as soon as the context
206.
by vb-8448 5mo ago
My experience is that in medium/big codebases even with single functions going with the xhigh is basically better from a user perspective (faster to get the result, and you can trust it) while going with lower models(e.g. sonnet instea
207.
by vb-8448 5mo ago
We are speaking about local AI , and having all this SOTA models basically for free is blocking the progress of local or independent third party setups.
208.
by vb-8448 5mo ago
> Use cloud models only when they’re genuinely necessary. The problem is that it's much easier to use the SOTA models (especially if they are subsidized) instead of spending time fixing the knobs with the local one. I just realized
209.
by vb-8448 5mo ago
> or do we LLM-build it from the ground-up? But in this case you don't spend tokens only on your workflows: you have to patch it constantly, perform vulnerability scans, check and adapt for law changes(eg. if you in europe: GDPR or
210.
by vb-8448 5mo ago
Let's take a SW business like a ticketing system. Do you think 100 enterprises with 1 bln of tokens are going to make a better product than specialized vendor with 100bln of tokens? For sure SW vendors and SAAS like "logo creator&

More ›