If you keep clicking down the rabbit hole, down at the bottom of their Methodology section, Waymo explains they're adjusting the data they're using to account for differences in human and Waymo route selection, variation by locale, etc.... based on peer-reviewed papers published by Waymo employees.

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwaymo.com%2Fsafety%2Fimpac...

If you dig through their methodology description what you find is that they're adjusting their numbers based on papers published by people who work for Waymo. And they also promise that they're counting all the accidents where there was any damage or anybody got hurt, because of course they report all of those, doing otherwise would be a crime! Or at least a rule violation. (From their phrasing, I get the impression that there are a large number of unreported, or at least uncounted here, "the Waymo vehicle bumped into something at very low speed and no damage was done" incidents.) And that they're still doing an unspecified amount of remote driving by humans. Which all boils down to "Self-driving car company says its self-driving cars are safer than human drivers!"

I get that the data cannot be usefully compared one to one with the data for human driver. Even if you had perfect reporting, Waymo simply isn't going to choose the same routes human drivers will; there will be stretches of road Waymo will likely never travel, just because of its business model. And there's probably not a lot of good research being done on these statistics (although I don't know that for a fact), so it makes a degree of sense that Waymo employees would be publishing. But claiming "we're using peer reviewed standards for our data adjustments" without a disclaimer that the writers work for Waymo is not a good look, nor does it encourage confidence on my part.

But one item I was not able to find, is where Waymo would fall in a ranking of human drivers, What happens if you take out the worst 1% or 5% of human drivers? How about if you take out incidents involving impaired drivers? It seems like these would be natural data sets, but I was not able to find them. (Maybe my search-fu and time is just insufficient.)

And since it will be friendly, we could call it, "super friend intelligence". That seems like it would be quite accurate.

I'm sure their AGI that's coming any day now can figure that out for them. /s

No need for that last one. The Trump Cabinet already has that covered.

Insane dictator is insane, says (and does) insane stuff.

The judges who aid and abet him, the Congress that fails to remove him, they're the bigger traitors here. Trump belongs in a managed care facility. They belong in prison. (And the "impeachment is not a priority" Democratic leadership aren't much better.)

The loss of meaning is intended.

Altman can't hype spicy autocomplete snake oil as a miracle machine if public discussion makes clear distinctions.

What, like one of the enlisted guys? Maybe an E-3?

Surely you're not suggesting anyone important be held responsible for their actions? This is the United States of America! /s

The current United States government says a lot of things, many of which are decoupled from reality. This statement was very likely issued solely because the occupant of the White House wanted to hear it. This "admission" or "acknowledgement" has no zero relevance to the question of whether or not the United States actually has on-orbit weapons.

Also also, we need an excuse for why we're not hitting the milestones I promised.

Other recording devices do not constantly record on their own, waiting for you to give them the instruction to do so after they do the actual recording. That is what it seems the spyWatch is doing.

This is illegal in many states. Apple is demonstrating the modern corporate position of "breaking the law is not illegal if a corporation decides it will make money doing it".

Once Trump is gone, the people who voted for Trump, and those who have enabled him in office, will all be unchanged.

This is just stage setting for moats to protect big, established players down the road.

Your cheap, local, open (and/or foreign) AI might be a dangerous threat to all humanity! Bill Gates says that you cannot in good conscience be allowed to own or operate such things, or at the very least there must be a level of regulation and taxation that is ruinous on the would-be small scale AI operator. Only the wise and experienced OpenAI and its equivalents - which by sheer coincidence have deep connections to Microsoft and Gates fellow billionaires - can be trusted to provide society with the massive benefits of AI. In return for government subsidies and a generous private profit of course. But don't worry! In the spirit of philanthropy, Gates will promise that some of the benefits of AI will trickle down to the masses.

I think we're going to get a new spin on this old line. The market *will* remain irrational, while *we* (the public, collectively and individually) go insolvent. But the finance sector and the billionaires will be kept whole, which is what they and their government (because it sure doesn't belong to us) want.

"Amazing but oversold" might well sum up the whole LLM-era.