I think LLMs may have a role in security posture, specifically flagging/identifying potential threats for human review. But a Zero Trust/Access Controls should be a HARD boundary, not an inconsistent one.

The problem we have right now is that there are some legitimately interesting ideas out there for things we could be using LLMs for, but we also have a ton of "I have a hammer, and everything looks like a nail" going on too.

[0] https://en.wikipedia.org/wiki/Export_of_cryptography_from_th...

1. Most states require special device certification by a "nationally recognized authority" that doesn't seem to fully exist yet. There ARE devices that would almost certainly GET certified, but aren't yet.

2. Ordering panels is a PITA. Most want to sell you either RV/Camping grade (overpriced), or in too large of quantities (e.g. 10x+). If you want to hit under 1,200 Watts meaning just 2x 600 Watt panels, well, nobody wants to ship that. Most popular panel vendors, don't want to sell single-panels, and you're looking at 500% markup for camping/RV style panels (with often worse performance, and proprietary connectors).

3. "Moveable" panel requirements. The laws frequently don't define what that means or what that looks like. It seems to want to essentially ban a small roof install, but stuff like ground-mounting (e.g. even just for the angle), or securing so they don't blow over, is ambiguous. Plus most laws don't exempt you from planning requirements, just utility requirements, so the local city/county can still fine you for your "install."

What I am saying is: I'm absolutely in love with plug-in solar, but unfortunately being an early adopter in the US is going to be annoying. I hope in ten years it will be almost turn-key, just buy a "kit" plug it in, and you're done.

Compare it to the Consumer Price Index:

https://fred.stlouisfed.org/series/CUUR0000SA0R

This may be an odd reference, but I remember this song and dance with Roomba Vs. Roborock. "Roborock stole our IP!!!" but the problem was that song was slightly believable in the very early days, then Roborock started releasing products far better than anything Roomba had, but the tune in the US never changed. Literally for years after Roborock's products were two generations better than Roomba, but somehow they "stole" that?

China has money, talent, and cheap electricity/superior infrastructure. The US's big AI built their products on stolen copyright material and social media scraping. China likely got a catchup boost via distilling, but recently a lot of their AI ventures have come fully online, and they've just throwing tons of compute/power at the problem, and we're starting to see legitimately competitive products without shortcuts/cheating.

But the US won't change its tune; and that's convenient because they don't really want to compete, but rather roll out protectionism and fear per normal.

So your account is now tied to a physical device; great, but the device is dead, or you own a dozen devices, now what? Each vendor has their own idea about what THIS means. Heck I have a couple that allow, max, a single Passkey at a time.

Perhaps, but in the last year China did build these:

https://www.navalnews.com/naval-news/2025/03/stairway-to-tai...

- Only one specific device can ever login (bad).

- It doesn't limit login to one specific device, therefore it does nothing.

Linking Passkeys to a physical device was always DoA. At least not without a way to enroll every device you own, and strong recovery strategies. But considering how inconsistent every company's Passkey implementation is (inc. many that only allow ONE TOTAL!), it is DoA.

The problem with "someone makes a request, code changes happen automatically, and all someone else has to do for that to be committed is mash approve" doesn't strike me as a way to create a maintainable code-base.

The changes may even work, and may technically fulfill the request, but the agent cannot know the design intent or process intent. Today there may be a programmer sitting between the request and approval, but what about tomorrow? Might it be a non-technical or barely technical middle manager?

It will cost you more than it saves to use smaller Chinese models to code; because of the repeated work. That has been slowly changing recently, but with much larger Chinese models, however those models are so expensive they're much more price-competitive iwth the US competition.

But for actually providing end-user AI features, particularly simpler ones, the US isn't even in contention. The costs and limitations just outright kill those features conceptually.

What's frustrating about that is that Microsoft has also gone out of their way to make it difficult to access the [legacy] System Properties (sysdm.cpl), while not fully reimplementing all the features into the Settings app. Including this one.

They've only been working on this 10+ years...

The title of this thread is "What AI did to stackoverflow in a graph." That's a narrative. At least before the mods change it.

This is revisionist history. People told SO that they were leaving for YEARS because of how incredibly toxic it had become. It was already giving outdated answers before ChatGPT shipped, because new questions/potentially updated answers were [Closed] [Dupe] immediately.

Their answer was essentially "We aren't a Q&A site, we're trying to be a knowledge base! So closing all questions on a Q&A-stylized site, and extremely abrasive moderation, is working as intended."

They entirely did this to themselves. The community was toxic, their policies were toxic, and they didn't listen when warned as such repeatedly - just doubled down.

The monitor only sends a unique device ID, everything else is handled by Windows.

Microsoft needs to intervene here, this cannot be a normal expectation for using their product.

You don't actually explain why or how these names are "easy to understand" just state that they simply are. That's great; to me, they aren't obvious or intuitive at all. May have well just start randomly pointing at dictionary words.

You don't, because that isn't something I proposed using for model naming.

I called them GPT-5.6-Max, GPT-5.6-Plus, and GPT-5.6-Fast. Reasoning levels are distinct from the model design itself, and the UI makes that clear.

Plus, using that same flawed argument this would be called GPT-5.6-Sol-Low or GPT-5.6-Luna-High which also makes no sense/is confusing. So that argument applies (or more accurately doesn't), no matter the model names.

I agree with them, Sol, Terra, and Luna are confusing names. They mean the same thing as GPT-5.6-Max, GPT-5.6-Plus, and GPT-5.6-Fast but require base knowledge for an analogy.

It feels like it was adding by the marketing department.