Once again, I am absolutely gobsmacked at just how clueless upper management at corporations have become.

How do these guys even manage to put their pants on the in morning. Do they have help for that?

Or more than one? I don't count M-X Doctor.

That's not the issue - insurance companies have no problem giving insurance to humans which have "individual learning capabilities."

The problem is that while insurance companies can extract compensation from humans - essentially they can coerce the humans to work "for them" to pay them back - they can't get compensation from the AI, and there isn't a good framework to get compensation from the AI developers.

The issue isn't that the AI learns - the issue is that our legal framework doesn't have a good way to get "compensation" from errant AI - because the AI itself doesn't care if it gets shut off. You can't hold the AI itself "liable" like you can a person. So until we change the liability framework, we'll be stuck. This isn't a technical problem, so much as a social one.

Part of self-driving cars being safer than people is that they don't drive "very close together" - they follow at the appropriate following distance and speed for the lines of sight. This alone accounts for a large portion of the improved safety performance of these systems.

Autonomous cars are actually very good at the driving task. Where they fail is when they encounter situations that are "abnormal" and I think the industry has a problem here; state of the art is either to train specifically for every situation (impossible in the limit) or to kick the decision to a human in a support center somewhere. Things like emergency vehicles, school buses, farmer markets, construction zones, etc.

The industry needs to figure out how to get their systems to "get confused, then reason their way out of a situation."

This long tail is where humans still do better - when there is some "new" scenario, like a particular combination of debris in the roadway, on a Tuesday, with gawkers, during a fireworks display. You will never be able to "train" your way out of that - you need the correct structural formulation for the problem space.

The big takeaway I have from their study is that their study areas - major metropolitan centers - is that cities have way, way worse than average incidents per vehicle mile travelled.

I mean the national average is only something like 1.25 fatalities per 100M miles, injuries only 75 per 100M miles, and "all" incidents like 140 per 100M miles. But the benchmark numbers on the linked websites range from 200 to almost 700 injury incidents per 100M miles in those cities.

Like, if you are using the average number, you can't even reduce the number of incidents by 841 over 271 million miles, because the expected value is only 203. You have to have a rate of at least 300 injury-incidents per 100 million miles to reduce by 841 at over 271 million miles. So really, their results are driven by San Francisco, Austin, and Atlanta.

Their confidence bands are also pretty low; for example they quote 2 serious or greater injuries in San Fran, which on 82 million miles there is a rate of "0.02 per million miles". But 2 injuries is not really enough to make a statistical inference.

Least of all because all the miles are with different versions of their software - you can't make any predictive claim.

Statistics never lie, but you can make them say whatever you want...

Adderal, duh.

(Well, in the quantities he *wants*, anyway.)

This is the prevailing dogma, but has recently been challenged from more recent observations. (In defense of the dogma, we have very aggressively tested for preferred rotations with local experiments involving high energy particles. If there *is* a preferred rotation, its effect is vanishingly small, to the point of being functionally indestinguishable from flat spacetime at local scales, to an astonishing level of precision. This strongly suggests 'sampling error' for JWST data that follows)

https://academic.oup.com/mnras...

Recent large area surveys by JWST suggests that there is a preferred rotational direction and axis to disparate galaxies. Such a finding is inconsistent with the presumption of widescale anisotropy. (But again, that's a very bold statement)

It may yet be ruled out through further observations as a sampling error, but the statement that there are *no* indications isotropy is incorrect. The mere existence of this survey says otherwise.

If there really is a universal preferred rotation, then a 'fuzzy', 'proximal' position could be given as a baricenter. It wouldnt be definitive; too much can happen as random events to compute an exact baricenter, but a statistically probable, proximal area could be defined with some confidence threshold.

Again, *assuming* the JWST readings arent just a sampling error. Which it probably is.

But we need instructions for toothpicks. AI robot flossing requires no instructions.

Wonko! Come back! We're sane now!

But WE'RE MICROSOFT, and we've decided that YOU WILL GO HERE today!

All our extremely cherry picked analytics data says it's what you really want, deep down, and you KNOW IT, so stop complaining! (Honestly, this is just like the whole "you got rid of our start menu" thing all over again-- Cant you people ACCEPT the new future of computing?!)

That "reality" bubble. You should check it out sometime.

Source: Guy who lives in and around immigrants and actually knows what he's talking about in real life.

Wikipedia was "banned" in schools because it's not a primary source. Just like how all other encyclopedias were banned from being listed as sources for papers. You're supposed to use the encyclopedia as a link to the primary source. Encyclopedias were just pre-Internet search engines for books, usually listing a little summary. Wikipedia being on the internet makes it convenient, but no different than book encyclopedias with regard to how papers should list their sources.

No, I think it's closer to teaching people how to drive on city streets and highways, possibly even on the "wrong"-side, when all they think they need is to drive on a race oval. Sure, oval track learning will get them the concepts of acceleration, braking, and turning with the wheel, but they won't learn actual vital skills they should have for maneuvering the car from the garage to the track.

Boss' Boss: "Why are we paying him to remote in when we can hire an international team to remote in for 24/7 redundant button clicking at half the price of one person?"

International team: "North Korea is willing to pay us to do the job we were hired to do. Sounds tempting..."

1) "smaller" = "Bigger" (in terms of sales, and modern utility. This is exactly what I meant, UV Light has a threshold over which the structures are smaller than the wavelengths you are trying to use to make them. this means you have to use some kind of mask technology, and you get strange quantum effects from the photons doing things photons do when they are restricted by tiny masks.

2) You are forgetting something here, to make a tortured inference that doesnt really fit anyway. Most of the technology used in a fab is either precision electronics (needed to control the energizing source's true positioning to efficiently control its beam), which you can do with less sophisticated chips than the ones you are making. The inverse creates a chicken and egg problem, and is nonsense. You dont need 2nm tech to make 2nm tech. You can make 2nm chips with 14nm ones though. The knowledge of how to make and run fab machines is not to be discredited-- Japan's big LED rise came from a single guy repeatedly rebuilding and tinkering with a vapor deposition machine at a university in the US, where he was treated like ass, because he didn;t have a pHD. DO NOT MAKE THAT MISTAKE TOO.

3) It is not a breakthrough. It's a milestone on independent progress that is occurring at a very quick pace. Again, you are proudly crowing about how big your old growth forest is, while discounting how quickly your competition is doing forestry work.