Firstly, requiring physical access is a massive barrier. You'd have to have a massively lucrative target to even risk trying to find what machine in a data center to (reading an article)...

Put in a device that sits between the memory and the CPU. Yea, nobody is going to notice somebody replacing all the memory in a machine with some random parts. Oh, of course that doesn't mess with the signaling at all and the BIOS will post perfectly with some janky ribbon cable setup.

And then, just sent the data via cellular internet from inside a metal box inside a metal box inside a big old building? Just hope nobody notices a random extra CAT 6 sticking out randomly from the box?

Look, research is all good, but security also lives in a practical context and I doubt Intel or AMD are really worried about this at all.

Sure, the tariffs are doing their work here, but it's not just that.

If you have to go to one and only company to make your products, you are at the mercy of what they want to charge you and they will extract as much profit out of you as possible. TSMC is almost in that position right now and Intel is the best hedge against that happening. So what if some of this is government pressure to keep advanced fabs in the US. Fine by me.

The real wrinkle here is when for demand for AI compute contracts (or just collapses) and there becomes a big glut in fab capacity. I understand I am saying when here just to be hopeful, but so be it.

Not that data ownership and control isn't a big issue that needs to be addressed. It is.

The issue is how social media has injected itself into our culture. So many people, without a single thought, document their day to day in the form of text, photos and videos with no thought to why and to what the consequences are. Because they have been trained to do so. To support levels of consumerism that are not sustainable in the long term.

The solution is sociological. We have to put social media and related technology at a proper distance. Banning access to smart phones in schools is a start. It needs to be coupled with a lot of media literacy, parental and governmental support to reduce and largely eliminate the usage of social platforms by those under 18.

And no, this isn't about unworkable laws for age verification. We have to break some fundamental patterns here. Because the impact on our overall health and functioning as a society as a whole is being negatively impacted to a degree we can't ignore.

The work structure isn't sustainable. Sure, China may have outlawed the practice just to look good to the rest of world, but it could also be because it will destroy your workforce. It will make them ill and will kill them prematurely.

Or, they get organized and some heads start rolling down the streets. Always a possibility.

Well, for a few people, they feel more empowered by taking other people's lives. It's the ultimate sense of control. A sense of power that is enabled by access to firearms without necessary background checks and ongoing certification and training.

Are their campaigns on how to recognize when you aren't able to use a firearm safely because of your emotional state? Mental health resources to call to help you de-escalate and manage your emotions better? Laws that would allow people to report their concerns to authorities so that intervention could occur?

Never mind resources that would help you with those things before you ever thought about getting a gun or getting access to one, because feelings are for snowflakes, I guess.

I mean, it's a good use of a cash from your hostage, err, loyal customer base when there isn't anything else around to suck the living life out of, err, bring new levers of enterprise stability to.

But you have to have somewhere to go when things deflate. In this case, you just have an amazing amount of hardware assets you can't repurpose to do anything outside of supercomputing and (knock on wood), they still have their own clusters.

It will deflate. The wow factor has worn off and the claims are getting more insistent with less and less actual data. Bigger models aren't leading to bigger gains, the slop is already poisoning the well. There are no noticeable real productivity gains, more and more actual revenue is just middlemen trying to get brought up while the money is still sloshing around.

When the biggest player in the space is part of a backroom deal to mandate usage of their product in a federal government agency, things have already gone off the rails. It just takes time for the whole thing to crash and to see what is left in the wreckage.

Instead of coasting on brand name recognition and making everything possible a yearly subscription to protect your margins. At this point, they deserve to get run over by Korean automakers and Chinese upstarts.

I agree with using tariffs to prevent market dumping by China, but you have to build something in response not just complain. So, go design and engineer a great EV. People like rising to a challenge, after all.

Well, it shouldn't be, but in the US, there is a lack of availability for therapeutic resources and pressure for doctors to just bill for medication consults (quick and pays well). This is despite many doctors getting into psychiatry because they like doing multi-modal treatment with therapy and medications as need, but always providing both or working with a team to provide both. And it's not just ADHD. Depression, there is huge pressure to prescribe when CBT based intervention could give a better outcome without having the real risks coming from SSRI tapers for those patients that don't need long term medication support. And for those that do need long term medication support, that and therapy always works better.

I get the pushback every time a study like this comes up, it raises the whole "simulant meds bad" pseudoscience despite the field agreeing that simulant medications can be a safe and effective intervention. But, medication should always be complemented by effective therapy to provide support and additional skillsets. There are worries about overdiagnosis, but again, that points to the need for effective resources. Psychiatry is perfectly capable of distinguishing why someone is struggling with school, but that takes time and insurance, parents and schools push for quicker interventions. something gives.

An AI company who is completely dependent on mining every bit of data out there, copyright be dammed had another company who also is completely dependent on mining every bit of data, copyright be dammed violate their terms of service?

How awful.

No, wait, that's completely expected. It's sad that the best we can hope for is these companies get shut down by the courts because LLMs aren't fair use.

But, hey, companies can't inflate their market value and stock prices by announcing sensible investment in labor and making them actually more productive, versus making existential threats to the workforce and hope it doesn't trigger the biggest labor movement ever seen. Which isn't a bad bet, given how much has been done to discredit organized labor as force for positive change (note, organized labor dovetails with unionized labor, but it isn't the same). Then again, if you work two jobs and still have to share an apartment, maybe the odds aren't as good as those large companies would like.

And I can't really argue with that premise set by the article. Windows 11 Pro is better, but increasingly not by much.

I get why Microsoft has to ride this AI bubble and is trying to push Copilot everywhere. The stock markets are fickle like that. And shareholders, blah, blah.

It is the cost of operating systems being essentially free, but having to support development *somehow*. Apple has captive hardware. Linux has tons of free will and some enterprise support.

Windows, it's pretty clear that OEM licensing fees are shrinking. But instead of stepping up and providing a better experience and making things opt-in, it's back to the same old marketing and sales of turn it on by default and see what happens.

You could sell plenty of Microsoft 365 subscriptions with just a few tweaks to Windows, OneDrive and Office to reduce the suck.

If OneDrive didn't sync anything by default and had a retention policy for synced files that were deleted that was reasonably sane, it would be way better.

Office, make Copilot an addon. Done.

Windows: Fix backups. FFS. New system imaging tool, file based backup (not OneDrive) that is sane (no, copy and paste is not sane) and make a good experience out of previous versions. Seesh.

This battle was waged. OOXML won. It's hacky, it's the default, there is probably better, but nobody cares enough to force the use of ODT. This is about working documents in organizations. The read only archives are all PDF.

If you want to displace those use cases, you need to have product that brings its own values and advantages to Office 365 users. No, being free isn't enough. Office and Windows is eighty bucks at most for a user per month. It doesn't take much to recoup that cost in organizations, which is why it is so common. The competition has to be good enough to charge a similar amount because it gives a similar ROI.

That is really hard, so best of luck.

Too many investors and users saw the vision around AI in iOS and they wanted it. And the backpedaling is hurting them, because they can't get there soon and may never get there.

That kind of thing has to just work. Text summarization? One two many gobbled messes and boom, that feature gets turned off.

Oddly, Microsoft isn't under similar pressure because, well, nobody expects them to do "awesome" with anything. If Copilot comes to mostly naught, bring in the Clippy 2.0 memes and they will move on.

Not that I care if they are using other companies models to ease costs. You can't inhale the internet, wave your hands about copyright and then complain "IP" when somebody uses your stuff in way you don't like.

If it takes more air out of the AI bubble, all the better, I say.

But, honestly, web standards are more and more rapidly becoming whatever gets put in the Chromium code base.

Which would be fine, if Chromium was meant to a be reference implementation. But it's not. It's used to keep people going to the web and Google for as much as possible. After all, need that ad revenue.

At this point, we are stuck with web applications on the desktop. I'd love a desktop version of things like Amazon, Facebook, Insta and so on. But, the story on Windows isn't compelling enough and enough people (well, really developers) use MacOS and Linux to make this impossible.

So, I am stuck with every single web site using a slightly different way of adding payment information, of managing accounts and profile information, for logging in and out, uploading files and so on. It's a mess.

Maybe Firefox can be a stopgap to things being even worse. I don't know.

I get it, schools are hardly in a position to fight against AI tools and all the marketing around them, and are rolling over (just asking people to cite the AI they used). That still misses the whole point.

The reason plagiarism is a fundamental issue isn't some abstract ethical point, it goes to learning. You don't learn by taking other's work and presenting it as your own. And using a chatbot is even worse for learning for that then finding, reading and copying actual work.

And coding is no more dead than it was when (insert tool here) claimed it would be every decade since the 70s.

So, in the end, the schools are devaluing themselves and the value they provide to students and the community. Higher education is useful because people *can fail* to get degrees.