You are right, they did get a dual input warning, but for some reason ignored it.

They did, but it's been a long time that that worked. Even longer since strings were literals.

There are good alternatives too now, like GOWIN. I know it's not always an option, you might need Xilinx features, you might be familiar with them already, but I find it's often worth the effort to move if you can.

As I recall one pilot held their stick back to keep the nose up the whole time, which doomed them. Airbus averages the inputs from the two pilots. Boeing produces a "dual inputs" warning.

Because he held back the whole time, the other pilot's efforts to level the aircraft and gain speed were ineffective. As you say, it was noticed at the last minute, but too late.

That always seemed like a very weird design choice to me. But also, the crew should have noticed sooner, and they should have made it clear who was flying and that the other pilot needed to let go of their controls.

Seems to be more a case of enough AI tokens and the source code, and all bugs become shallow.

Presumably Microsoft has Copilot doing the same for Windows, and Apple has some AI working on MacOS and iOS, and we know Google has been using Gemini AI for Android.

They just quietly fix stuff before it becomes public knowledge, but Linux is open source so can't really do that.

Perhaps, but what a great job the unions did negotiating even this much. A massive bonus, nearly 8 times annual salary, and even if the stock price does crash eventually, those workers still did extremely well out of it.

On the upside, AI lets anyone make a movie.
On the downside, AI lets anyone make a movie.

Including people who have terrible taste in plot, style, and everything else.

There's some genuinely good stuff out there - Gossip Goblin's work for example. But this is....

I'll just say, there's far better things that one could have spent half a million dollars on...

I'd think so given it appears to have a monochrome low resolution LCD screen and controls that while I'm sure are functional, are far from "gaming grade".

The most interesting part will be what radios it has. The CC1101 in the original Flipper Zero is a great chip. I started using it long ago for work and soon realized it is extremely flexible.

Even there though, I think all the commercial RISC-V SoC suppliers use binary blobs and undocumented peripherals.

Unless they have competition, they will go for wireless "broadband" instead of upgrading to fibre. It's shitty but it's cheap, and they don't care if the service is any good when you have no other choice.

Ironically it was exercise that screwed up some of my joints, due to undiagnosed health issues. It's hard to know what is best to do.

Stressing about it is probably worse than the damage a lot of this stuff is doing. Plus I need coffee, life isn't worth living without it. I'm not joking.

How are you defining "statistical inferences" as distinct from "logical inferences"? If you're defining fuzzy logic (e.g. not necessarily yes-or-no answers but allowing for ambiguity in conclusions), then we can agree conceptually, but your choice of wording is, I have to say, weird if so.

My understanding is that LLMs are built on a foundation of ANNs, and that indeed the backpropagation used to train ANNs is a statistical process;

Two responses. One, that's discussing individual-neuron scale processes rather than collective processes; and this was a discussion about inference, not training. Human neurons also learn by error minimization (Hebbian learning). But this does not describe the macroscopic processes that result from said minimization.

* During training, neurons develop into classifiers that detect superpositions of concepts that collectively follow the same activation process. Individual neurons weight their input space and subdivide it by a fuzzy hyperplane to achieve a classification result.

* In subsequent layers, said input space is formed from a weighted combination of the previous layer's classification; thus, the superpositions of questions being formed are more complex, as are the classification results.

* In a LLM, this iterates for dozens of layers, gaining complexity at each layer, to form each FFN

* The initial input space to a FFN is a latent (conceptual representation), as is the output; the FFNs, in result, function as classifier-generators; they detect combinations of concepts in the input space, and output the causally-resultant concepts into the output space

* FFNs alternate with attention layers dozens to hundreds of times in order to process the information, each layer building on the results of the previous one.

The word to describe that is not "statistics". It's "logic".

In a LLM, the first few layers focus on disambiguation. If there's a token for "bank", is this about a riverbank, a financial bank, banking a plane, etc? As the layers progress, it starts building up first simple circuits, and then progressively more complex circuits - you might get a circuit that detects "talking like MAGA", or "off-by-one programming errors", or whatnot. In the late layers, you have the general conclusions reached - for example, if it were "The capitol of the state that contains America's fourth-largest metro area is...", you've already had FFNs detect the concepts of fourth-largest metro area and encoded Dallas-Forth Worth, and then later taken that and encoded "Texas", and then finally encoding "Austin". And then in the final couple layers you converge back toward linguistic space.

Anthropic has done some great work on this with attribution graph probes and the like; you can detect what circuits are firing, and on what things those circuits fire, and ramp them up or down to see how it modifies the output. They very much work through long chains of logical inferences.

I use every style imaginable, including photos, in my tests. Same result every time.

One time I even did it with a Calvin and Hobbes comic, pretending than an AI made it. Responses included things like "The illustration also looks like shit and barely makes sense. Hope that helps.", "God damn this sucks so bad", "This also fucking sucks", and "The only punchline here is casual, pointless cruelty. if you think this is funny then you're literally a psychopath."

It's true that if it works as intended, Starship would be a fantastic tool. But aside from having doubts about that, it doesn't preclude others from getting them on a similar timescale, just a different way.

Blue Origin demonstrates that. Panned for being "behind" SpaceX, but when they fly stuff it tends to work and suddenly they caught up. The Chinese are the same, and they aren't the only other people working in this area. That said, the rate at which some of the private Chinese outfits have been advancing is really impressive. Okay, they had second mover advantage, but they had that with EVs and batteries and renewables too, and are now way ahead...