Is there a button for upvoting an article? The only thing wrong with it is the implication that there might be legitimate places for AI.

There may well be. After the last AI boom (remember Expert Systems?) there were a few useful tools around, but only a few.

Just because certain people are trying to ram AI anywhere doesn't mean that it is automatically without use. Blindly vilifying AI is as much an error as bashing it in everywhere.

"Blindly vilifying AI is as much an error as bashing it in everywhere."

But more likely to be right most of the time.

I think you mean "More likely to be right than any given AI response".

My Linux workstation is remarkably AI free. I have a Docker with Python and my own LLM's, but they can be deleted with a single line of code.

I control the OS, not some corporation. Windows is the other way around and you'll pay them monthly to access your own data and like it.

A similar thing to this happens in the world of cycling. Oval chainrings are one of these ideas that seem obviously useful but which aren't actually much cop when you start to use them. This one has popped up in the cycling world at least three times and generally gets reinvented, hyped like crazy and in the face of complete disinterest forgotten about for another few decades.

The same will happen with LLM AI systems, save that they mostly won't resurge again ever. Yes, they have uses same as blockchain has uses, but nowhere near as many as the proponents of the things think that they have.

> Whatever legitimate places AI has

The only valid use of AI is to separate the gullible from their money. Anyone pushing AI at this point is either a fool or a fraudster. Or, more likely, both.

Convince me otherwise.

I think it's pretty clear - the "Other channels" is management.

The same Management who cannot be seen to have pissed away multiple billions chasing a fad which anyone with half a brain could see was going to go the Way of Web3.0 or the dotcom bubble in short order...

"Other Channels" could conceivably be the intelligence community. Remember when Microsoft _allegedly_ put an NSA DLL in Windows?

"The same Management who cannot be seen to have pissed away multiple billions chasing a fad ..." It is said that Bill Gates made that selfsame point to management in 2019 "you’re going to burn this billion dollars... (by investing in OpenAI) "

What's worse is insisting that a user-level application or feature is so intrinsic to the OS that it cannot be removed.

What's even worse, like the IE situation mentioned in the article, is intentionally making an application intrinsic to the OS so you can later argue that it cannot be removed when people start asking questions about your abuse of market dominance.

Internet Explorer, Windows Defender (W10 onwards), Teams (at least early W11 versions), and now copilot. This shouldn't surprise anyone.

They also tried with the MS Store, which is nearly impossible to get rid of without breaking the OS (the latest 25H2 finally has a policy, but Enterprise only)

I notice my MS365 homepage having a bigger and bigger copilot textbox, and it's taking more and more clicks to get to the place i need.

My prediction? They'll make Copilot an integral part of the UI, argue they cannot remove it, and probably get away with it too.

Having learned nothing from Clippy or Cortana.

Maybe they should just be banned from using anything in the Windows UI that starts with a C?

Perhaps they may benefit from a more real-world example, by learning the difference between a tool and a tradesperson.

AI can have uses as the former, but when it tries to be the latter, it's (and we are) doomed.

The agentic platform might (in an ideal world) act like a Type 2 hypervisor - situated between the OS and the active user desktop (and application suite) if it's turned on and active. That seems like a reasonable way to do it - transparently intercepting a user action and turning it into a series of integrated application and system calls. And like a hypervisor, it ought to allow multiple agents to run simultaneously, if that's the way Microsoft wants to structure their AI. If turned off, the active user desktop would run normally (natively) over Windows without AI. The user could also switch on the fly between agentic AI and non-agentic (no AI) desktop that way, too.

I figure that could improve reliability, useability and stability as well as respect user choice.

Otherwise, like any good Type 2 hypervisor, it should be completely removable.

Anyhow, it's all academic to me - I use Linux.

"We hear you"... Corporate talk for "Stop whining and be content with what you get".

Dog's got to bark, a mule's got to bray

Soldiers must fight and preachеrs must pray

And children, I guess, must get their own way

The minute that you say no

Why did the kids put beans in their ears?

No one can hear with beans in their ears

After a while the reason appears

They did it 'cause we said no

... The song "Never Say No" from the musical "The Fantastiks" https://genius.com/Hugh-thomas-actor-and-william-larsen-actor-never-say-no-lyrics

Am I saying that tech management resembles petulant children with beans in their ears? Well, yes. I might be suggesting that.

I am on Linux so it does not matter to me, but this is going to be a popcorn moment.

If the Agentic OS is forced upon existing users, then will this slow down the PC due to the resources required by the AI ?

We all know that Microsoft will f*ck it up, one way or another, so it is going to be fun to see the fallout.

Maybe, just maybe, next year will be the year of Linux on the desktop....

I'd be right there with you were it not for being forced to use W11 at work :(

My sister is forced to use W11 at work. She is also required to disable and avoid LLMs in every possible way for client privacy. If Microsoft make an LLM compulsory at the OS level she may well be required to upgrade to something without AI. Would your work place be happy to send every keystroke and mouse click to Microsoft where it will be used to train an AI? If they are, pick a few articles on the internet showing how to get training data back out of an LLM and see if they really like the idea of sending their trade secrets to competitors.

Whilst this an excellent article in the main, there's a fundamental lack of understanding of the nature of Big Tech (I originally wrote that as "Bug Tech" which is an apposite Freudian slip) in the statement that "Microsoft is in the business of engineering".

Microsoft is very definitely NOT in the business of engineering. It could be argued that it was up until 1979 but since the 80s it has existed only to make money. Nothing wrong with that in a capitalist system but they're no different from all the others such as Apple, Meta and Google. Their foundation myths that they're somehow hippy types different from conventional corporates are just that - myths intended to distract attention from the fact that they are cash machines and we're the metaphorical cows being milked.

It is bad enough when dealing with interactive chatbots that can be ignored by intelligent people, but I do not want my operating system hallucinating on me when managing system resources.