retains access to the AI startup's technology until 2032, including models that achieve AGI
Exactly how do they envision an autocomplete gaining sentience?
It hasn't been "autocomplete" in a long time. Sure, there's a training step based on a corpus of Human language, and the autoregressive process outputs a single token at a time, but reinforcement learning trains specific behaviors beyond merely completing a sentence.
Besides, the best way to write something indistinguishable from what a Human might write is to, well, "think" like a Human.
performance or yield, combined...
Why should this not be a violation of the Americans With Disabilities Act?
My eyes cannot deal with the "flyspeck 3" fonts used by smart-phones, and as I result I cannot use one. Can I get the statutory $15K per violation specified by this US law?
I suspect he doesn't know about - or thinks most people don't know about - ad blockers, and so offering a browser that has more relevant ads will attract people who hate the irrelevant ads but figure they're unavoidable.
Obviously, for those of us who are smart about adblocking, who almost never see ads we don't agree to see, it's stupid.
But shockingly, there are probably plenty of people who will happily sign up for that "improved" experience.
This means that the climate-modeler assumptions about the lifetime of free CO2 in the atmosphere are out to lunch.
The whole world has realized that they need to start air-gapping databases
I've worked at government contractors that had real air-gaps for things like their databases, but that does not seem to be the norm for the rest of the world. How would ordinary businesses make use of their databases if they are not network accessible under any circumstances, printed reports? Some sort of unidirectional transmission? What sort of data ingress are they using?
I ask this because I have been involved in the transfer of data in highly regulated, air-gapped systems, and they are incredibly expensive. Are you really indicating that true air-gap databases will be ubiquitous (or at least commonplace) in the forseeable future?
Not surprising - given the accident description, the pilots dramatically exceeded the structural limits of the airplane by dramatically exceeding the posted limits of safe operation.
Here's the report of the event:
https://asn.flightsafety.org/a...
It reads to me as if the (co)pilot deliberately amplified the Dutch roll motion (which is a simple lateral/directional dynamic characteristic of aerodynamics of ALL aircraft, whether or not it's operationally significant) until reaching such a large motion (probably a combination of simultaneous roll and yaw rate) that it exceeded the strength of the engine pylons.
In particular, it's possible to overstress the aircraft by putting in large OPPOSING inputs in an attempt to stop to the Dutch roll motion too quickly. It's quite possible that if they'd simply gone hands/feet off the controls, it would have damped out naturally with no damage. The failure of AA587 is an example of this: the pilot's large rudder input against the yaw motion (which was due to a yaw excursion due to unexpected turbulence) is exactly what caused the tail to snap.
Nearly every claim in this post is factually wrong. This person apparently knows nearly nothing about aircraft flying characteristics, other than buzzwords.
I am a degreed aerospace engineer working in flight test for over 30 years and I have tested large commercial/passenger-class aircraft, including deliberately-induced Dutch rolls for test purposes. I have sat in the cockpit behind pilots executing these Dutch roll maneuvers intentionally. I personally joined a flight test team that had a crash a few years before, due to a Dutch roll event during flight test in the early 1990s. I also edited the US Naval Test Pilot School handbook FTM-103 "Fixed-Wing Stability and Control Theory and Flight Test Techniques" in 2019-2021. So I'm working with definitive expertise acknowledged in my field.
Some facts.
1) The proper term is "Dutch roll", uppercase, not lowercase, just like "American flag" not "american flag."
2) Dutch roll has absolutely NOTHING to do with the wings alternately stalling. Zero. Nada. Its cause is more subtle and would take a few pages to explain; go look up section 5.2.2.3 of the USNTPS FTM-103 flight test manual (which I edited) if you want the math. A wing stall MIGHT occur AS A RESULT of Dutch roll if you cause such a very large Dutch roll at a very low speed, but I've never seen that happen in hundreds of flight test events, some of which were done with me in the airplane. And even if the wing DOES stall, it's not a big deal most of the time; I've also done many stall tests. We're careful and we know what will happen, but most stalls are immediately recoverable (because all airplanes are carefully designed to recover quickly and gracefully from an inadvertent stall).
3) There is no "triplet" input to "get out of" Dutch roll. It is a natural oscillatory motion and will persist until a "yaw damper" is engaged to counter it automatically; it's almost impossible to manually damp out because it will simply recur naturally. Even if you put in a complicated input to damp it out, it'll start again just due to small gusts. No pilot wants to spend all their time fighting Dutch roll; that's why aircraft have yaw damping systems.
4) Dutch roll absolutely CAN shear off the vertical tail if it becomes large enough. Look up the crash of American Airlines AA587 in 2001. And I personally worked on a program that crashed a Navy S-3B test airplane in 1991 from Dutch roll testing when the vertical tail failed due to bad test technique (deliberately overdriving the Dutch roll mode beyond the limits of the vertical tail strength, due to miscalculation of the tail strength limits). I personally have on my desk the control stick that was recovered by wreckage divers from the bottom of the Chesapeake Bay to remind me of that failure. That crash still informs Navy flight testing practices today.
5) But in normal operation with a properly-functioning control system and absent extreme pilot inputs, Dutch roll will never become large enough to cause a failure; all aircraft are designed with sufficient stability to not reach this point without a control system input (either deliberate or due to a hardover rudder input). It is, however often a nuisance residual motion which can be annoying or nauseating.
6) Dutch rolls not exactly a "very slow" oscillation. Slow, but not VERY slow. In most large aircraft, its period is about 5-6 seconds per cycle. Smaller aircraft have faster oscillations, maybe 2-3 seconds. It's easy to observe, and quite annoying.
See my root-level post here with more information about Dutch roll in general, and the actual issues in this event.
I have a few observations about this incident, because I see some false information in this thread.
I am a degreed aerospace engineer working in flight test for over 30 years and I have tested large commercial/passenger-class aircraft, including deliberately-induced Dutch rolls for test purposes. I have sat in the cockpit behind pilots executing these Dutch roll maneuvers intentionally. I also edited the US Naval Test Pilot School handbook FTM-103 "Fixed-Wing Stability and Control Theory and Flight Test Techniques" in 2019-2021. So I'm working with definitive expertise acknowledged in my field.
Some facts.
First, the proper term is "Dutch roll", uppercase not lowercase, just like "American flag" not "american flag." Basic respect for Holland and all that.
Dutch roll, according to the US Naval Test Pilot School handbook FTM-103 "Fixed-Wing Stability and Control Theory and Flight Test Techniques", is defined as a second-order oscillatory lateral/directional mode of oscillation, sometimes referred to as a nuisance or annoyance motion. It is characterized by an oscillation back and forth between roll and yaw (sideslip); if you look out the side of the aircraft you'll see the wingtip trace a small circle or oval path. Nearly every airplane will exhibit Dutch roll - it's baked into the aerodynamics - the only question being how susceptible it is, and how well it naturally is damped.
So every airplane ever flown is deliberately tested for lateral/directional stability including deliberately inducing Dutch roll to check for its damping characteristics: will it naturally die out or is the "yaw damper" needed to reduce it, and how fast is it reduced by that system.
Dutch roll testing is actually one of the more benign types of flight tests. We are careful to avoid exceeding the sideslip limit of the airplane, so we build up to the larger test points, but it's generally quite safe.
You can see a sample plot of relatively representative Dutch roll motion here: https://www.researchgate.net/p...
My read of the article and other information I can find about this incident is that it wasn't a Dutch roll that *caused* the problem. "Dutch roll" is simply a natural mode of oscillation present in any airplane, and won't lead to any aircraft damage unless something else went badly wrong. In this incident, it's likely that the rudder power control unit (PCU) had a "hardover" or oscillatory failure and drove the tail to swing sideways (either once or oscillatory) far enough to cause physical damage to the tail (the FAA preliminary report says "substantial" damage), which I presume manifested as popped rivets and visible sheet metal buckling (I've personally seen these before after flight test events that went a bit too far). So the failure was not Dutch roll itself. Presumably the pilots immediately turned off the Dutch roll damping system (the "yaw damper") and maybe even the rudder PCU itself. The resulting residual motion after such a failure, without the yaw damper reducing the oscillation, might be characterized by a sustained yaw/roll oscillation which WOULD be a Dutch roll mode of oscillation, but that was the effect, NOT the cause.
Dutch roll is not inherently dangerous. What *is* dangerous is when it is excited (by pilot input, by control system failure, or by wind gusts) and becomes large enough to cause structural damage, usually to the vertical tail due to side loads on the tail. The crash of American Airlines AA587 in 2001 is an example of what happens when the vertical tail fails due to overstress; the pilot encountered an wind-gust disturbance and used too much rudder to try to correct the motion, and literally sheared off the vertical tail. So modern aircraft include rudder limiting systems to prevent large inputs at high speeds.
In this incident this week, the rudder PCU apparently malfunctioned and caused a huge yaw input, leading to (probably) vertical tail damage.
AFTER the real failure and any damage that occurred, because of the failure of the PCU, it's likely that the rudder couldn't properly damp the Dutch roll during the rest of the flight, so there would also have been a sustained but small oscillation which can cause people to get nauseous or worse due to motion sickness.
If you want to learn more about Dutch roll, here is an older public-domain copy of FTM-103 (the new version I edited is not being released publicly) and you'll find the relevant information in section 5.2.2.2.3.
https://usntpsalumni.wildapric...
Finally, is this a symptom of Boeing design = bad? No, I don't think so. Every aircraft has a ton of parts which must be well maintained. Servos (PCUs) fail due to wear and tear. Aircraft maintenance practices are designed to spot failing or worn parts such as PCUs and replace them before this happens, but sometimes it still happens. That has nothing to do with the fundamental design. And as I have tried to convey above, Dutch roll is inherent to every airplane. The only question is how good are the systems to limit and damp it out, and how good is the servo design to not cause it. Since this and thousands of other 737 aircraft generally have no problems like this, it's not a design flaw; it's a worn-out-parts flaw. So if anything, I'd pin this as most likely due to maintenance not catching a worn part, but maybe just bad luck.
Oh, so there you are!
