Not always. It’s better to say that “military grade” means “very specific standards”.
That also happens to translate to cheap pretty often, but that’s because someone in 1943 wrote the specification for The Standard Tube Sock and ordered 7,500,000,000 pairs to be delivered over the next century. Since it’s precisely specified you know exactly how much you can skimp.
Add that to the trillion pages of compatability requirements and you’re basically never updating anything.
Like how switching lightbulb brands at a nuclear power plant is hard because you need documentation demonstrating that the plastic in the led won’t become a hazard in a radioactive incident.

Do they require disclosure of what font the developer used for coding the patch? No one thinks it matters, but no reason not the be transparent, right?

The reason not to require disclosure is because in their opinion it doesn’t matter any more than the font matters. If the person reviewing the code can’t tell it’s AI, and the code stands on its own, then it stands on its own. If they can tell then it doesn’t meet code standards anyway.
A perspective that I don’t disagree with is that “disclosure” can be a shield for the developer to hide behind. “Anything weird wasn’t me being dumb, just the AI being wonky”, and that’s writing an excuse for subpar code. You submitted it, it’s your code, and if you can’t justify some dumb bit in it then you deserve the shame without the help of a face saving “it was the AI”.

Also: people trying to make a point out of whatever motivation.
Need a POC for a research paper? Do it in Linux.
Need an implementation to get investor funding? Do it in Linux.
Need to pressure a large commercial operating system to implement a security model as a state security agency? Do it in Linux! It’ll only take a decade for them to start to trust it, but the company will be pressured by a free alternative have an implicitly endorsed security system!

Your statement ignores the meaning of AI or the decades of work that have gone into the field.

They’re not sentient. They’re not sapient. They’re not conscious, nor do they have feelings, volition, desire, or anything that can be described as an internal life. They can’t even think in they sense that anyone means by it.

None of those things are intelligence though.

An LLM posesses the ability to respond to external stimuli in a way that is meaningful. They also have the ability to learn, which is the act of a system transforming information or stimuli in a way that is applied to later stimuli.

My air conditioner learns and is intelligent. It measures the environment, takes actions based on that stimulus, and when I make adjustments it records when I like different settings and applies that to the temperatures it’s recorded in the past to adjust early. An LLM may be more sophisticated than my air conditioner but it’s in the same category.

The mistake is thinking that AI in the real world is what AI is on TV.

I went and read the commit descriptions.

Previously it was using whatever cursor the theme had loaded and just scaled it up. Now it references the full set of resources for the theme and picks the appropriate one for the size.

So they went with the most correct third option: defer to the standard set by a more authoritative source. :)

People in charge of the company I work for are actively delusional.

We had a four hour mandatory meeting to go over our product strategy. The high level vision. For the engineers. No rest breaks. It was announced there would be a test on the materials. Not as a joke.
The top question in the Qa section afterwards was “what is our product strategy?”.
The meeting had roughly 10k attendees.

“AI agents are going to become so pervasive that we’ll need to put a computer on every desk in every office. I’m serious: have you tried to buy a Mac mini lately? They’re just not available”. “Datacenters in space”.

We think they were announcing a new product dashboard, but we’re not sure.

A huge amount of freakout about how AI will make everything different and better that most people on the ground don’t agree with, but we’re mandated to reorganize stuff according to the theory of how it should work in their vision. Teams that change and shift every few weeks. No fixed goals in the week to week timescales. Less planning, more demos, and development goals dictated by higher ups AI agents.

When it all shits the bed they won’t be the ones getting fired. They’ll be the ones collecting the same paycheck and talking about how “limited restructuring” is hard for everyone, but market forces dictate.

I’ve never been under the impression that my work makes the world a better place. I’ve always sold a product that goes to companies that do stuff that might help someone else help people.
It’s someone else having a goddamened coke bender on 50k people’s livelihoods with no consequences that has me down.

I won’t disagree with what you said, but I’ll disagree that it’s the same point I was making.

My point was that there isn’t a technical solution to traffic congestion. You need less fluid or a bigger pipe. What you correctly mention is why a bigger pipe won’t work for humans when it does for water.

And out of pure pedantry, I’d disagree purely because my analogy said “non-newtownian liquid” - those don’t expand. :P it’s more like a hyperactive toddler manically jamming silly putty into a PVC pipe.

Traffic congestion doesn’t happen because of lack of coordination. Coordination prevents collisions. Our success rate could be a lot better but the system does pretty well with coordination limited to blinking lights, a noise that says “fuck off”, the mutual knowledge that we all got the same training and the private knowledge that everyone else on the road is a suicidal idiot who slept through the training.

Congestion is a physical property of the road network. There’s a physical limit to the number of cars you can push through a road in a given time period. Drive the cars too fast and they fly off the road when it bends, or cars slam into each other when a deer tries to commit suicide.

Traffic is well modeled as an incompressible slightly non-newtownian fluid, like runny oobleck. If you apply more pressure/increase vehicle volume the fluid starts to bind up as it becomes harder for the cars to maneuver without collision, regardless of coordination.

You reduce traffic by taking cars off the road, or scheduling utilization. Self driving cars won’t help, and we can do both of those without them. Long term they’re likely to make collisions worse because humans don’t know what they’re going to do, so the primary mode we have of coordination, “the theory of mind” will break down. Right now they’re so uncommon that you assume any car is a human who reacts and prioritizes like you.

Tear down the freeways and replace them with trains.

Almost certainly.

They don’t have to, since it’s not a part of the system that it requires them to hold onto the information, but they all invariably hold onto the information to keep training on it, which includes your personal information.

They are likely not using for selling to other people or anything since that’s not their business model at the moment, but it’s probably finding it’s way into the next version in some capacity.

Absolutely no idea. I almost logged on when typing the previous reply to see if they were still offering me the “thin blue line commemorative gun holster” they were the last time I was on, but I realized I really didn’t want to.

I feel like I see it being used as a common enough way for events to share details though, so it probably isn’t just older demographics.

It’s some of that, on some platforms. Other platforms also get paid to promote posts or advertisement. Depending on who pays the most, you can see different things based on your demographics.
An important thing to remember is that they will show you something in the money making category. If they know what you want, they make more if they show you something that’s likely to be effective. If they don’t know what you want, they make more if they show you something targeting your demographic. If they’re fuzzy on everything, they show you something.

Which political geographic is more likely to target Facebook? Which one is more likely to spend demographic money to try to court younger white men from a low to middle income bracket?

Your phone fell off the dashboard phone stand, you caught it and set it in your lap.

It happens every couple of decades with AI. Since it’s a broader field than most people think, we have a pretty long cycle of a new development looking exciting, people getting way too excited and optimistic, the development being exactly what it was promised to be, and then people getting disappointed and avoiding anything with the AI label. Then we decide that because we’re used to this new thing, it can be used in stuff as was originally appropriate but it no longer qualifies as AI, because “that’s not AI, it’s just ___”.

Yup, that’s the precedent bit I mentioned. They’re both valid ways of deciding ambiguity (a judge decides in this situation and other judges aren’t obligated to decide the same way, or lawyers debate the ambiguity, a jury decides, and courts strive for consistency), but they’re fundamentally different and slightly bewildering to people from the other model.

Purely an answer to the question "I mean, how different could they really be?”: fundamentally different in an almost unrecognizable way.

I don’t know how the German appeals system works, but there is a lot of room for difference.

A particular reading spree once caused me to learn that the UK, a modern civilized country, didn’t have what we would call a supreme court until 2009.
Their laws aren’t codified. We have a big book o’ laws, and we pass bills that modify the book. If it’s not in the book it’s not a law. They pass bills that are the laws. This sounds really similar until you consider that “the law” is a collection of every act of parliament going back nearly a thousand years, many of which cancel out others. Oh, and that extends to the concept of a “constitution”.

Some quick searching shows that Germany uses a fundamentally different legal model that views our big book o’ laws as unstructured because courts have a binding say in interpretation of the law. It seems that this regional court can be appealed, and also that their courts don’t use precedent like ours do, so an appeal is more like a second opinion than an escalation.
Judges are less referee and more investigator, so you can claim that the judge made a mistake with their decision, which is appeal.

Yes, I will just take your word for it over the word of the original people involved.

You keep talking about DARPA, when they’re not the same organization that backed ARPANET. Arpanet came before laws were passed saying DARPA could only fund projects directly related to defense.

DARPA was originally ARPA. They were under the department of defense but their project scope wasn’t limited to defense projects. The reorganization that rebranded the agency as DARPA and made it defense focused ostensibly saw the non-defense oriented moonshot project responsibility transfer to the NSF, although the funding shift wasn’t proportional.
The order of creation isn’t exactly relevant to how responsibilities have shifted.

It’s kinda like how, for the longest time, presidential security was handled by the Treasury department. It wasn’t because presidential security was considered a financial matter, but because that’s where it fit.

https://www.darpa.mil/news/features/arpanet

Secure communications and information-sharing between geographically dispersed research facilities were among the ARPANET’s original goals.

From your link to the arpanet wiki:

Building on the ideas of J. C. R. Licklider, Bob Taylor initiated the ARPANET project in 1966 to enable resource sharing between remote computers.

Sutherland and Taylor continued their interest in creating the network, in part, to allow ARPA-sponsored researchers at various corporate and academic locales to utilize computers provided by ARPA, and, in part, to quickly distribute new software and other computer science results.

There’s a big difference between ARPA funded labs and general university usage.

I’m not sure why it would matter that you worked for them in the early 90s. That doesn’t exactly give you a privileged insight into the creation of ARPANET.

The Internet that you’re posting on was built on top of a military network intended to provide redundant communication in the event of a global thermonuclear war.

Responding to this part alone: that’s not actually true.
The intent of arpanet, the direct predecessor to the Internet, was to make it easier for universities to use high powered computer resources located at national laboratories, as well as making it easier to distribute software updates. The person who initially pushed for it’s creation wanted “an electronic commons open to all, 'the main and essential medium of informational interaction for governments, institutions, corporations, and individuals '”. They secured funding for the initial computer science labratories, os research that underpin everything, and the foundation for the “INTERgalactic NETwork”.

Arpa was, at the time, the advanced research project agency. They were under the DoD, but they filled a role closer to the NSF today.

In designing the system they referenced work done by people who were studying robust communication networks. At the time that meant the phone system and nuclear weapons. The research, however, was applicable to any unstable network, and so had particular interest to them because computers had terrible reliability and they wanted to not have to call people if they discovered they had turned off a computer halfway between New York and LA.

The closest thing it has to a cold war military objective is to help us win the research race and spite the Soviets. It can withstand a nuclear attack, but that’s just because that’s the easiest way to make it survive a farmer with a backhoe accidentally hitting a wire.

The thing is they’re currently trying to sell as a business oriented tool. They’re not going to make money off of individuals.
Google is positioned to come closest because they’re already an advertising company.

If you think their focus isn’t businesses then you’re not paying attention to their strategy. The pressure to drive profitability is increasing as their business customers are reporting that investment in AI capabilities isn’t converting into measurable financial returns. That’s the type of news that makes investors wary.

If you operate at a loss, you need to be providing a value to your customers that you can leverage. You need more than high interest, you also need demonstrable utility.

There have also been plenty of times that a new technology just … Didn’t pan out. This specifically happens with AI technology, and we even have a term for it: https://en.wikipedia.org/wiki/AI_winter The tech won’t go away, it’ll just be market infeasible for a while until it’s no longer called AI and is just a feature in some other product.

Take your comment and apply it to someone marketing “spell check as a service”.

This is capitalism working exactly as intended. People with capital are using it to guide businesses to make them more money.

The mistake is thinking that capitalism is motivated by a healthy economy, when the theory is that if everyone is aggressively selfish then the economy will naturally become healthy and efficient.

The people making money are counting on making their money as investors keep pumping more in. The investors are all aiming to be the last one to sell before the crash. Russian roulette venture capitalism. (In some cases they think the economy will tolerate 1-2 companies and they’re aiming to buy a controlling interest in a company worth 2% current value after the correction, and in others they just have so much money that a few billion is a minor bet - BlackRock comes to mind, with more than $10 trillion to invest)

If you look at 2008, we had a similar-ish situation with a major portion of the economy being invested in mortgage based investments.