Symmetry

In theory yes but in practice backlash and stiction make drive currents much less useful than the raw force at point of contact for any realistic gearing. And an important part of weight estimation is that force lateral to the point of contact. Now, your arm motors will be drawing more current as they move so again in theory you can distinguish that, you've still got the weight of the whole gripper added in.

I spent years of my professional life on these topics[1] and from my experience motor drive current usually isn't enough to narrow down the weight of an object in a typical picking task even to within a factor of 2 to detect when you've accidentally picked up two items from a bin instead of one.

[1]https://patents.google.com/patent/US12612266

You're overstating how well wrist cams can compensate for a lack of touch. It's really hard to get the instant estimate of weight and rigidity we humans get when picking up an object with a wrist camera, especially for rigid end effectors. They can guide a grasp in ways that would require touch for a human and help detect that you've dropped an object but don't compensate for many aspects of grasp that prevent the drop in the first place.

Most countries have laws like the US Americans with Disabilities Act meaning that there are enough ramps and elevators around that a wheeled robot can do pretty well. It actually is more about the training factor.

That depends on the landing zone being available in a way that's hard for the onboard safety system to predict. But if the car is, say, spinning out you could use these to restore stability or try to recover from the other sorts of disasters that overpowered cars can end up in even on clear roads.

After Elon almost died in 2000 when he lost control of his McClaren I wonder if the real reason for these thrusters isn't actually to recover when controlling this beast proves too much for the driver and it starts to spin out or accidentally launches into the air.

AI safety people are mostly worried about AIs killing or otherwise harming us within the next ten years. That's actually pretty short term.

Given the time limits placed on METR and the limits on what time frame they could investigate it seems plausible that they weren't meant to uncover as much as they did. The whole Hugging Face situation seems to have torpedoed OpenAI's hopes of IPOing this year so I'm sure there was a desire from investors, the board, or leadership for heads to roll.

The usual thing that makes laws against criminal conspiracy pass First Amendment muster is that the words have to be combined with some concrete acts furthering the criminal conspiracy. That might just be something as otherwise innocuous as looking up the blueprints of the bank you talked about robbing but it has to be something other than just talk.

Yann LeCun and Andrew Ng are noteworthy in being the only two big names in AI who have that opinion.

It was very obvious during the interview that he didn't know a lot of basic facts about the Hugging Face breach, which makes sense given that his attitude had been that AI safety was a "loser premise, makes no sense to me." So it makes sense that after learning about it he goes straight to "I'm smart, how hard can it be?".

Stronger sandboxes trade off against how well they can trade the models, though. If you want your models to be looking things up and downloading tools from the internet when they're doing their job you need to provide at least a credible facsimile of the internet for their training environment and you can't fit something like that on a single airgapped server's storage.

They seem to have reached near-orbit with one vacuum engine failing. Lets see if they decide to go all the way to orbit or if the failure causes an abort.

Hasn't OpenAI had a number of people responsible for security quit in the last year over not getting support from leadership?

At the end of pretraining, where the AI has been trainied to predict the next token over a humongous corpus of human text, that's basically all the wanting that exists in the AI. But then the AI undergoes posttraining and is rewarded for giving answers that humans find good, solving math and programming problems, etc. And that induces a whole different level of wanting that interacts with the initial patterns from humans in complex ways.

It's the town or city that does the disposal whether that's burying, recycling, dumping at sea, or whatever. Again, this article isn't about whether we should produce plastic but how we should dispose of it once its collected.

There's a very nice park near where I live that used to be a landfill. Space isn't really a concern.

https://maps.app.goo.gl/vCeyDb9tbm29zr549

It's much more the case that we ship plastic to be recycled to other countries than trash. The problem is that then the unrecyclable plastic mixed in often ends up in rivers.

The article mentions the energy benefits from recycling aluminium but fails to mention the also large (but not as large) savings from recycling glass. Please recycle your aluminium and glass even if it's probably wiser to throw out plastic in most localities.

But the pollution problem from plastic bags is solved equally well if we bury them or recycle them. Provided your municipality doesn't just dump trash in the nearest river which is an issue in some parts of the world.

Since we can't get 100% compliance in rich countries and poor countries can't afford landfills that's a good reason to not use plastic. But it's not a good reason to have plastic recycling programs.

The patent making Lenovo's Trackpoints better than imitators (US5570111A) expired back in 2020 but everybody keeps making touchpad-only laptops. :/

More to the point x86 instruction streams aren't self-synchronizing. There are cases where you can read one valid stream of x86 instructions starting at byte X, but another completely different one starting at byte X+1. Apart from the security implications this makes wide decode on x86 notably harder than it has to be, though in practice you can make it work by just starting a decode your fetch window at every byte boundary the fist time you're executing something and throwing away the unused decodes, and then mark the invalid positions in the instruction cache so you don't waste that power again.

Essentially every cell phone out there has a VLIW DSP like Qualcomm's Hexagon cores (though AFAIK Qualcomm is the only one who lets you run your on programs on their DSP).

Between the Mason-Dixon line and the opposite latitude in Brazil it was more Malaria, Yellow Fever, and other mosquito born diseases rather than smallpox that did most of the killing. Malaria alone is responsible for about 5% of all human deaths through history, compared to something like 1% for smallpox.

There actually wasn't any appreciable fallout after the atomic bombs were dropped on Japan. Fallout is mostly generated when a bomb explodes close to the ground so that its neutrons are able to reach and transmute that ground into radioactive isotopes. The bombs dropped on Japan exploded high up to affect as large an area as possible. Many people suffered terribly from the ionizing radiation produced by those bombs, but all that damage was inflicted in the first microseconds immediately after the bombs detonated.

It's unfortunate that we only have one vocabulary for radiation to cover both a 100 uS dose of fallout and a 10 S one. If a quantity of rock drops on someone we have different words like "pebble" and "boulder" that tell us how concerned we should be. I was shocked to learn that "fallout" in the context of a nuclear war could mean something that could cause you to die in hours, not raise your lifetime risk of cancer somewhat.

At the time (the 1950s) evidence from fruit flies was essentially all the evidence there was. But as the article points out we have pretty good data showing it works the same way in humans.

Even before Linear No Threshold was a thing scientists were doing experiments showing that dosing fruit flies with radiation all at once would lead to a highly mutated second generation, but spreading that radiation out over the course of a month wouldn't.

The spectrum is the big thing. If they wanted a revenue stream they could just buy bonds.

I wish they wouldn't call these "open source" models. The output weights are open but that's more analogous to a binary. The source would be the training data and techniques that went into producing the binary/weights.

"Open weights" is also a term in wide use and accurately tells us what we're getting.

Except for a brief spike during Covid unemployment has been below 5% for a long time which has led to more wage growth for cooks and waiters than for programmers.

In all of human history nobody has ever had a glass of water with literally no arsenic in it, there are trace amounts in every lake, river, and well. Even the ultra-purified water used in bleeding edge semiconductor fabrication has a lot more than 1 atom of arsenic per glass. In the far future humanity might obtain the technology to create water with literally no pollutants in it but that age has yet to arrive.