ein0p

51 ms·

961.
by ein0p 2y ago
The US is a major user of Israeli software and hardware though, Mossad notwithstanding. Seems imprudent from the “national security” perspective
962.
by ein0p 2y ago
Ask Snowden.
963.
by ein0p 2y ago
> Saying "no" is easy and safe Not in my experience. Saying no to something major when others don’t see a problem can easily be career-ending.
964.
by ein0p 2y ago
The US invaded 7 different countries and killed upwards of a million civilians by some estimates in this century alone. Should the same measures be applied to it, or is it “different” and I “don’t understand”. How about Israel?
965.
by ein0p 2y ago
If we’re going that far, wouldn’t one then logically assume that all US software has US backdoors? What does that mean for other countries?
966.
by ein0p 2y ago
How to lie with statistics, election year edition.
967.
by ein0p 2y ago
Why does such a simple site need any of those services in the first place?
968.
by ein0p 2y ago
With the Congress as rotten and corrupt as it is today, good luck trying to “rein in” anything of this magnitude. I’m sure the bribes will be quite a bit fatter though if they pretend to be interested in solving this problem.
969.
by ein0p 2y ago
I'd be OK with that, too, if we adjusted the bet for inflation. This is, in a way, similar to fusion. We're at a point where we managed to ignite plasma for a few milliseconds. Predictions of when we're going to be able to ge
970.
by ein0p 2y ago
Be that as it may, do you disagree with anything concrete from this paper?
971.
by ein0p 2y ago
It is in moments like these that I wish I wasn’t anonymous on here and could bet a 6 figure sum on AGI not happening in then next 10 years, which is how I define “foreseeable future”.
972.
by ein0p 2y ago
Apparently not. Or maybe they are heavily incentivized by the hype cycle. I'll repeat one more time: none of the currently known approaches are going to get us to AGI. Some may end up being useful for it, but large chunks of what we th
973.
by ein0p 2y ago
I am in the field. The consensus is made up by a few loudmouths. No serious front line researcher I know believes we’re anywhere near AGI, or will be in the foreseeable future.
974.
by ein0p 2y ago
We the HN readers do not need it, but for 90% of the rest of the world's population the interaction model with computers today for tasks more complex than Instagram is akin to casting magical spells, and it's been like that for de
975.
by ein0p 2y ago
That is why, btw, canned foods are either acidic, or sterilized under pressure at way above 212F - to prevent botulism
976.
by ein0p 2y ago
I was half expecting to find a real page there.
977.
by ein0p 2y ago
GPT4 by the way can generate simple OpenSCAD parts if you can describe them verbally. Even if it doesn’t get it exactly right, the result is a good starting point.
978.
by ein0p 2y ago
Under a “do not even look at it” license I’m sure. /s
979.
by ein0p 2y ago
I have a VisionFive2 board with the same SoC, and this is going to really disappoint if you expect laptop grade performance from it. Don't get me wrong, it's impressive for what it is (the fist such SoC that can comfortably and in
980.
by ein0p 2y ago
All outputs of an LLM are generated by LLM. LLMs today can and do use external data sources. Applied to humans, what you're saying is like saying that it's not humans who provide references because they copy the bibtex for them fr
981.
by ein0p 2y ago
Sure, but it can semantically query a vector DB on the side and use the results to generate a "grounded" response.
982.
by ein0p 2y ago
I’ve read they already offer rather steep tax incentives. I think what they actually need are higher wages, so one of the parents wouldn’t have to work for a few years, or at all. And that’s easier said than done, especially with their cur
983.
by ein0p 2y ago
No, humans legit just suck at associative memory for textual data. This happens to be one of the few things LLMs excel at. LLMs suck at sample efficiency, they do not model most types of cognition, they suck at multimodality (so far), in-co
984.
by ein0p 2y ago
LLMs can provide references. There’s no limitation on that. Even GPT4 includes references sometimes when it deems them beneficial.
985.
by ein0p 2y ago
My point exactly. The Homo Sapiens baseline is laughably poor in this regard.
986.
by ein0p 2y ago
Nope, that’s not how it works. Those references aren’t generated in such systems, they are retrieved. They might not provide references to all the sources, of course, same as humans.
987.
by ein0p 2y ago
Belief is a posterior distribution over a combination of priors and inputs. That’s literally the only thing LLMs currently do.
988.
by ein0p 2y ago
There are models that can provide references if you’d like
989.
by ein0p 2y ago
Neither do humans. You write what you believe. It is intractable for you to confirm every fact you believe in
990.
by ein0p 2y ago
Fabrication implies malicious intent or at least intentional deception. LLMs don’t have any “intent”.

More ›