Symmetry

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.