The Aztecs in particular were kind of uniquely terrible, both for their own citizens and for every oppressed pseudo-vassal-state around them. It's one of those weird accidents of history that Spanish colonizers were able to step into the power vacuum after the fall of Tenochtitlan and have at least some people genuinely think 'yes, this is better than the last boss'.
Cortez came to Tenochtitlan as an 'ambassador' at the head of an army of 200,000 angry neighbors of the Aztecs, who had realized pretty fast that even a few of these 'gun' things would be really useful for cracking the city's structural resistance to sieges.
Would be interesting how you'd steal it, it's on the moment you have it, emitting its location... maybe you put a blindfold over the camera/walk into a faraday cage then power it down/wipe the flash.
From the beginning they know who you are
Would be interesting people start hijacking humanoid robots, little microwave EMP device (not sure if that would work) and then grab it/reprogram it.
The problem with US cities is that they're not dense enough. Most of the US has spent the past half-centry actively making new high-density construction illegal or incredibly expensive, so everything is operating within the bounds of 1970s-80s construction being reused over and over again because it was grandfathered in.
People being forced by economic forces to move into uncomfortable and unaffordable dense cities. This in turn creates disadvantaged underclass with no hopes for a better future. And even European social safety nets can only do so much.
While just hours away from dense cities, the apartments are often literally free. With copious space and easy access to basic services.
This results in rising crime. The downward trend that started in 90-s had been reversed. And the crimes of despair, mainly drug-related crimes, are rising faster than violent crimes.
That ad gave me a visceral shudder of revulsion, not so much for the specific functionality on display as for the timing, which absolutely could not have been accidental. They might as well have just put 'and we're working on automatic alerts for ICE!' in the ad.
"Administrative subpoenas" have always been bullshit that mostly rely on there being no penalty for companies that hand over user information to anyone with a badge and then justify it with a five-hundred-page TOS document.
Google, among most other tech companies, deny portions of administrative warrants. Here's a story about someone who was stressed out about their notification by Google (spoiler, Google decided to deny the government's request)
edit: It appears that this outcome is an outlier and most admin warrants are honored. It is unfortunate to see the Washington Post decline in reliability like this.
Google does not provide those products (not in the US, as far as I am aware), but they are a money transmitter in the same vein as Square/Block, Stripe, and Venmo [0]. They won't be directly subject to the Bank Secrecy Act, but they partner with the major payment networks (who have their own rules and their own partner programs with banks) as part of Google Pay and customer payment profiles.
But I don't think this matters much for this case, as DHS is not investigating financial crimes. This is about what discretion Google has to comply with administrative warrants, which is not settled law and isn't clearly spelled out in their own policy.
I just looked it up, and money transmitters are included in the Banking Secrecy Act as "Money Services Businesses". So yes, they have KYC obligations in the sense that they know where you are moving your money and are obligated to tell investigators.
Unfortunately, KYC is used for much more than just financial crimes, and the precedent to comply is much more firmly established.
There is a case to be made that administrative subpoenas can be good. They save taxpayers money, they speed up investigations, and they free up the court for more important matters.
As with all things though, these agencies should not be self-regulated without civilian and judicial oversight.
I'm a little sad that they talk about counterfactuals in the simulations, but then don't show any examples of even a single sharknado or giant loop-de-loop.
Also manages to sustain itself on it's own weird brand of whales, a handful of disgruntled users with enough money to just keep buying accounts using random characters as a username just to get immediately banned after their first post. Some taking a dump in the middle of your living room isn't so bad if they are paying your rent and you can just kick them back out.
Complicated-enough LLMs also are aboslutely doing a lot more than "just trying to predict the next word", as Anthropic's papers investigating the internals of trained models show - there's a lot more decision-making going on than that.
> Complicated-enough LLMs also are aboslutely doing a lot more than "just trying to predict the next word", as Anthropic's papers investigating the internals of trained models show - there's a lot more decision-making going on than that.
Are there newer changes that are actually doing prediction of tokens out of order or such, or are this a case of immense internal model state tracking but still using it to drive the prediction of a next token, one at a time?
(Wrapped in a variety of tooling/prompts/meta-prompts to further shape what sorts of paragraphs are produced compared to ye olden days of the gpt3 chat completion api.)
Spotlight works great for me but the index can occasionally become corrupted. There’s a set of commands you can run to force a rebuild if you’re having issues. I’ve had to do this a few times over the years. This is another place where it’d be nice if Apple had a GUI to manage things like this.
reply