qayxc

25 ms·

301.
by qayxc 4y ago
> What is incredible is that it get’s this far. It can compute but not quite correctly yet. That's a conjecture on your part. The ability to compute is quite binary - either it can compute or can't. Humans often make mistakes w
302.
by qayxc 4y ago
> Google search absolutely does hallucinate completely fictitious results. It's called SEO spam. I wouldn't categorise that as "hallucinating fictitious results" - the algorithms still only returns existing results.
303.
by qayxc 4y ago
The problem is that the LLM is just that - a language model. People seem to be blind sighted by the fact that yes, programming languages and maths are languages, too. So the model is astonishingly good at transforming human language into co
304.
by qayxc 4y ago
Took me a while, but apparently sometimes you get a message saying you died from falling into a pit. Calling it "a game" is a bit hyperbolic, but the fact an AI model generated this from just a few prompts is still impressive.
305.
by qayxc 4y ago
NVIDIA invested billions in their software infrastructure over the past 15 years (CUDA was first released in 2007) and basically brute forced their way into academia back in 2008, sponsoring labs around the world to get GPGPU going. Both mi
306.
by qayxc 4y ago
> I think Starlink is doing pretty good still. Well, atm., yes. From the very beginning I thought that's it's a very risky idea to focus on end users, though. Ships, airplanes, mining operations, emergency services, etc. - that
307.
by qayxc 4y ago
Well, the methodology is flawed, or at least it doesn't actually reflect what it says on the tin. It's "consumer-initiated", i.e. what they are really measuring is which contracts people are using, not which speeds are
308.
by qayxc 4y ago
> They are justified only up to the point where Apple starts scanning your personal photos and then the EU joins in requiring everyone else to do that too? Interesting view. There was a time, not too long ago, when photos were developed
309.
by qayxc 4y ago
The line couldn't get any blurrier. The Swastika is a symbol that's been in use for thousands of years in Asia, Africa, the Americas and Europe. Its association with mass murder is so very recent, it's almost ridiculous that
310.
by qayxc 4y ago
Oh my, where to start... This little article shows everything wrong with software devs these days :) First of all, the implementations aren't doing the same thing, which is bad for a proper comparison; but others pointed that out alrea
311.
by qayxc 4y ago
Even worse: imagine the whole internet being flooded with AI generated regurgitations of older art styles. The next generations of AI models would then use this as training data, creating a feedback-loop that gets us nowhere fast.
312.
by qayxc 4y ago
Do you use any special hardware (like FPGAs) to mitigate the increase in computational cost or do you rely on standard hardware? I'm very interested in FHE in the context of machine learning models without requiring access to unencrypt
313.
by qayxc 4y ago
It can't do that. It can try to do that and it might work well in some cases. It is definitely incapable of doing that for any code you throw at it. It also cannot generate comprehensive unit tests - it can can generate unit te
314.
by qayxc 4y ago
> Github Copilot solved my business problem by itself just as I would've done. Is that real-world enough and the solution generalized enough? No, it isn't. Co-Pilot is unable to provide a rationalisation for the generated code
315.
by qayxc 4y ago
So once again we are talking about mice here and “We need to do this study in humans and confirm whether the same pathway operates in the same way.” Also in order to have any effect, you need to actually create a pathway for the bacteria
316.
by qayxc 4y ago
But micro-scale von-Neumann probes wouldn't be building civilisations either - they'd be l'art pour l'art - building more Dyson Swarms for the sake of building more Dyson Swarms. The paper also only considers microscopic
317.
by qayxc 4y ago
FTL would defeat causality and even "just" controlling gravity would be basically equivalent to god-like powers (it would mean assuming direct control over spacetime itself). Controlling gravity would negate the need for colonisat
318.
by qayxc 4y ago
> We don't have a view of what exactly is happening right now across the other side of the galaxy We don't have a full view of the other side of the galaxy. A lot of the galaxy is invisible to us (blocked by dust and stars) and
319.
by qayxc 4y ago
There's an even simpler one: if you leave your stellar neighbourhood, you're creating unreachable isolated islands, not "galactic empires". The whole notion of "expansion" is such a weird 19th century concept t
320.
by qayxc 4y ago
The real question is whether any "filters" are really needed. Spacetime alone is sufficient to explain the apparent absence of detectable technological civilisations. There could be a Dyson Swarm in the making right now , but eve
321.
by qayxc 4y ago
The very concept of a "Galactic Empire" is flawed as soon as you take scale and spacetime into account. The first major obstacle to the concept is distance. It'd take centuries to get even to the nearest off-world colonies an
322.
by qayxc 4y ago
> And of course I have no idea if the maths check out :) The galaxy is about 90,000 ly across so a time dilation of about 0.004 is required to get to the 40 yr figure. Doing the Lorentz transformation results in a required speed of 299,7
323.
by qayxc 4y ago
There's a difference between a natural process being harnessed using technology and claiming a breakthrough when there's extensive research and working devices that are are further ahead in terms of TRL already. But I guess you kn
324.
by qayxc 4y ago
I fail to see how this was "revolutionary" even back in 2020. The principle has been known since the Apollo-era when von Braun and his team discovered it by accident (rotating detonations destroyed the F1-engine's combustion
325.
by qayxc 4y ago
Don't confuse satellite launches with crewed launches. Losing a satellite due to a failed launch is no big deal, risking the lives of potentially hundreds of passengers is a different story entirely.
326.
by qayxc 4y ago
And don't forget the required pressure suits. You can't just to use simple masks like in airplanes, since cabin pressure loss at heights beyond 25km would make breathing using just oxygen masks impossible.
327.
by qayxc 4y ago
A "short boat ride" you say... Current exclusion zones and logistical requirements suggest otherwise. Realistically, P2P using giant rockets is a nothing but a pipe dream. Just straight forward things like weather proof this. The
328.
by qayxc 4y ago
Depends. If (and that's a mighty big "if") the fuel situation could be sorted out in terms of being able to produce it in a sustainable way, the footprint would be quite minimal compared most other modes of transportation. Yo
329.
by qayxc 4y ago
Agreed, I used to do that 15 years ago and I wouldn't do it again. The most miserable part isn't even the flight itself - it's the hassle at the airports.
330.
by qayxc 4y ago
You know what, just for shits and giggles I actually DID open an introductory textbook on abstract algebra, just in case my undergraduate degree in maths failed me. Here's what the first chapter says about addition ('+') in Z

More ›