Posts by C.Carr 35 publicly visible posts • joined 17 Nov 2016
Some models can pick the next appropriate image in a sequence, by figuring out the rule. Doing it close to human level requires a lot of very expensive inference compute.
People in these comments really have no idea what they're talking about.
Will the UK then ban the side loading of apps on Android devices? Will they ban VPNs? If they don't want anyone using user-friendly, E2EE messaging apps, that's what they'll have to do. ... but then I suppose they'll also have to ban flashing custom ROMs on Android devices ... and if you keep going they're going to have to ban GitHub ... and if you go even farther they'll have to ban semiconductors ... then the one-time-pad technique ... then writing implements ... ... and eventually math.
Not to completely dismiss the possibility of genuine psychological harm, but I wonder what these people thought the job was. They seem ... fragile to me, and certainly not suited to such work.
Short of outright horrors like snuff films and child porn, I'm pretty certain I wouldn't be bothered -- and, honest, I'm not a socio/psychopath. I was, however, born in the 70s, and so perhaps psychological put together a little differently than the kiddos these days.
If Putin were really that concerned about the expansion of NATO, he wouldn't have commenced an invasion that anyone with two neurons to run together could have foreseen leading to Sweden and Finland joining the organization.
Great job, Vlad, your border with NATO has increased by 1,340 km.
Putin knew Ukraine was never going to invade Russia. He knew Ukraine was never going to invade Russia.
Starlink sats are in a low enough orbit that they passively deorbit in a fairly short period of time -- like 5 years, IIRC, without station keeping.
Satellites above 600km are a much larger concern
I guess you can review the environmental impact of all those launches, but there is no environmental impact to review in, you know, space.
You're misunderstanding what this was. They set the booster down in the water at around 2 or 3 km/hr, in a vertical position -- which is exactly what they were trying to do.
It was a test flight. They intended to expend the booster. Eventually (apparently in the near term) they are going to attempt to set the booster down on arms extending from the launch tower. Whether that's attempted on the next test flight probably has a lot to do with just how much booster descent control they had, and how accurately they placed the water landing.
By the way, every single other US rocket manufacturer dumps their booster in the ocean -- IIRC, Rocket Lab may have recovered and refurbished one. Everyone else throws them away.
That hinge mechanism seems to be amazingly robust. I was sure that flap was going to tear off. I was very surprised to see it still actuating right before splash down.
Flight control software must be very good as well, to compensate for those missing chunks, being able to pull off the flip despite screwed up aerodynamics from the holes.
Glad to see that Starliner did manage to dock. I know a crew capsule isn't trivial, but Starship-level difficult it is not.
If you're finding exploits in the ChatGPT web app, or whatnot, sure, hackers -- but if we're talking about probing for "bias, hallucinations, and jailbreaks," with a natural language interface, they should get a bunch of academic linguists, maybe particularly people with applied linguistics degrees.
Welcome, Wumao. Y'all are indeed getting better.
The issue isn't merely that kids spend a lot of time on it, it's that they spend a lot of time on it *and* the CCP can do whatever they want with that attention. Bytedance has zero ability to refuse Beijing. Chinese government control over Bytedance is unparalleled in Europe or the USA (although the CCP has banned US social media, search, online tools, etc, anyway).
If tomorrow an app out of Sweden took American youths by storm, and stole all the attention away from TikTok, no one in the US government would give two sharts.
Human language is the product of embodied things that walk around on the surface of a big spheroid, at the bottom of a gravity well. Words point to things in the world, as well as at each other in a network.
You can't fully comprehend language by only examining relationships between words. A system that understands human language must be able to hear and see and pick things up, and feel the inertial mass of objects smacking it upside the head.
To state the obvious, the labels need to consist in more than their complex relationships with other mere labels. The actual things need to be represented by sensory data, and the AI system needs to be located, discretely, in 3D space --- that is, if we want a system to actually *know* what things are.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
