Microsoft have promised/threatened (delete as you see fit!) to bring AI into File Explorer. Please whatever gods anyone prays to may Opus remain free of any such monstrosity!
2 Likes
It's funny because if Microsoft's "agentic AI" actually worked like they claimed then they wouldn't need to add AI functionality to individual programs like File Explorer; the system-wide AI should be able to control everything itself.
I am not so sure on the AI topic anymore.. I was very sceptic and "anti" at first, but AI got me an answer recently, which I was searching for two weeks on the internet up and down with no result. That kind of convinced me, that there is value in this technology (if used correctly o).
In the meantime I hand more and more task over the "the robot", it is very handy if you don't need to enter the browser and search bits and pieces for yourself anymore. The AI agent does it all, convert files, syntax, ask math questions, produce code and it even reads screenshots and error messages you "paste" into it.
Today I handed a tedious refactoring task over to the coding agent and I just clicked "Confirm" after it did it's thing. It's a miracle to me how this works, but I am amazed each time about what it can do for you.
So, I can actually imagine some kind of AI type of thing to manage files. There are lot's of things you can write scripts and plugins for, but if the computer would sort and sync my files, pictures, downloads on its own and also rename these files to my liking automatically.. I mean why not!? o) I can't keep up with these tasks by myself (too many things collected over the years) so any help is appreciated.
Think of the MP3 collection you have. Now you tell the AI the pattern you like your files to be named, how the tags should be arranged and that each album cover with a low resolution is to be swapped with a hires one. This is a task which would take weeks and months if done by hand and is probably some minutes only, in which you can sit back and watch it happen.
I still see danger in using things you don't know how they work though and what exactly will happen with society, humanity and the internet, if everything is tied to a corporate AI tech of some kind, it's kind of scary, but refactoring source code and things, it works! o)
Getting a quick 100-200 lines of code, some kind of kickstart or prototype for new programs and ideas also works quite well. You can dive into XAML and WPF for some weeks with your own brain to create that little GUI application or just sit with the coding agent for 30 minutes and you are done with something you can work on from now on. If you have a bug in a method, hand it over to the AI, it will find the error for you within some seconds, it's crazy and I only use these coding agents now and then, I did not really dive into the topic or tried other engines than Copilot yet.
If you want that .NET and WPF application for Windows ported over to Linux using Python and Qt, no problem, it's more or less another command to the agent and you have that little application working on Linux - mind blowing. You still need to know what you want and also need to "see" when the coding agent is getting it wrong, so you still need to know your craft to some "deep" degree to make this work. People who never wrote a little script on their own will not get far win programming with that AI thing (currently at least, not sure where things end up).
I don't know.. this "AI" and me, it will probably end up in a love and hate relationship. It's probably like the regular "knife", you can do harm with it or use it to your own advantage, you have the responsibility.
I would also try to make that AI run locally at some point, just as I do not use any subscriptions or third party email / cloud providers. We should not "buy into" more big tech for now, because losing independency (completely) is probably not worth it, but that does not render the actual "robot brain" technology useless, using it correctly is the key I guess.
Oops, long text! o)
Sure it can have its uses, but I don't want it poking around my files - or doing anything, really - that I haven't explicitly asked it to. Microsoft has made it pretty clear though that they intend making Windows more and more an agentic OS. Can't say I'm thrilled about that.
I agree, but whether there is an AI agent on the computer or not, Windows can poke around in your files anyway - and it might have done so since forever, nobody knows, right? o)
So, is another "Service" on the computer more harmful to your privacy? I can hardly see that, there are already a zillion things in the Windows background you don't know what they are doing. The general idea though, to not expose yourself unnecessarily is right and valid of course.
I would still wait some years from now, until I would really consider to let anything AI to run over a larger amount of collected files in "full auto" mode. We will need to see how things turn out, there will be security and privacy analyses as well. If Microsoft is sending big chunks of your private data around, will not happen unnoticed.
And.. If you don't run Windows Home, you probably will have plenty of options to turn unwanted Windows features of. It has always been this way and it will work for professional / enterprise editions of Windows in the future as well (I assume at least). Most businesses care for their data and privacy as well, that's why you can turn off telemetry in Windows Enterprise completely and officially by using the policy editor.
I don't panic for any of this, the "Windows 10 Pro" I still use right now has been stripped down to an absolutely quiet work station. No forced reboots, no unwanted notifications, no auto-installation of applications or any advertisement. It's just rips through what I throw at it and "the Windows" does not get in the way. The "modern" Windows settings are horrible though, no question and no fix for that yet.
You basically have to keep using Windows anyway, if you don't want to move back in time to stone age with a Linux setup. It's a question of preference, for productivity I think Windows is the only choice for now. If you value privacy more, then yes, Linux is an option, but you will need to leave behind a lot of "goodies" from the Windows world.
I don't see me using Linux for quite some time, at least a decade or two! Unless they start to raise the bar noticeably within the next few years, but that's probably not gonna happen. The "diversity" in the Linux land is holding it back, endless forks, no standards, no common goal and lots of ideology mixed in.. o)
Wow, you have provided an amazing, balanced take on current AI technology.
Personally, I'm an AI optimist 
Thanks, I did not see the usefulness for "AI" at first eiter, but I changed my mind because of the positive results I got.
In the end, "AI" is just another algorithm, we should not concentrate on the marketing and buzz wording. We already adopted awesome things like JPG-compression, video codecs, syntax highlighting in the source code editor, speech recognition, raytracing and mind blowing 3D gaming and graphics. All of this is math, physics and algorithms.
"AI" is not different, it is just another information technology advance and as a computer person, you can hardly reject a clever IT algorithm, can you? o)
A local search engine like "Everything" is a nice invention. The google web search on the other hand, yeah.. it is the same nice invention, wrapped into lots of issues with privacy, tracking and even censorship, but that does not render the invention useless. It's just a bad example of using it correctly or a good example of using it incorrectly.. o)
Ahh, Yes that helps.
Been trying to understand the new December 10 Australian law here.
Can you explain? o)
One which produces endless wrong results (because it can’t say “I don’t know”, because it doesn’t really know anything), destroys the planet and commits copyright theft on an industrial scale while making it impossible for most people to buy a GPU or even memory, which makes it harder for people to use the algorithm even if they want to, which they do because the algorithm also made everyone forget how to think.
Other than that, it’s great, yeah.
2 Likes
Consider this too:
AI requires the output of a small nuclear rector (or equivalent) to produce its questionable output... the human brain (which built it, inter alia) can get by on a cup of coffee and a Mars bar! 
Hello! o)
I'm not going to be a lawyer for an algorithm here, but some things to consider..
- I think it can say "I don't know", kind of. Last week , when asking about an issue I had on a Win2022 Server, it gave some options to try, then settled for "well, it might be a bug", which is plausible.
- "destroys the planet": I guess you mean energy consumption? The ever growing energy consumption on planet earth is not going to be reversed. More and more energy being used, is a sign of technological progress and wealth. If energy can be "harvested" more easily, it will be used. It's like a big hard drive or a fast CPU, if there is space or calculation power left, it will be used at some point.
- Prices on hardware: You all are old enough to see the transition we did from a C64 with 1 Megahertz to some kind of Ryzen 32 core CPU at 5 Gigahertz. This is 5.000 times faster by 32, which is 160.000 times faster within 40 years. Now you want to tell me this algorithm won't run easily on your own hardware at some point?
- That copyright theft: Yes, that's problematic, but nothing new and not the fault of the technology itself. As I said (I guess), you need to differentiate the actual technology from how it's used by (greedy or bad) people. Just because Google is crawling the web and offering a search, you would not reject a search function in general, right?
- Regarding People who forget to think when using"AI": There is a quote out there, like this: "The internet makes the dumb people dumber and the smart people smarter." So, you can use your computer and GPU to play silly 3D games or run physics calculations for weather forecasts on it to aid humanity. What you are doing with technology is in your hands, right?
- The mars bar and cup of coffee vs. nuclear plant theory. I think we are past that already. Humanity won't go back to paper and pen, when there is spread sheet calculation, which allows to calculate so much more in less time, no way. o)
That said, I am also quite cautioned when it comes to big tech companies abusing some of this. That danger is not new though and the solution to this is knowledge and choice, not rejecting or forbidding the technology. A smartphone is a nice device, what makes it dangerous is not the technology, it is too much trust combined with basically no knowledge about what is going on inside of it.
So, my approach is: Make the people smarter! o)
There is a chance this won't work though, that's why I vote for the right to a "full analog life" (as an emergency fall back, even though I appreciate sensible digital solutions). You can do things great and correctly on the digital ground, but you can also mess up big time.
A nice sunday everyone.. o)
The point about mars bars and coffee vs a nuclear reactor isn't to make an objection to the amount of energy used by AI per se (though I think that is a valid objection that shouldn't be brushed aside as easily as you do) but rather to ask the question: is this even the right approach? AI needs this power because its learning methods are a kind of brute-force thing - this is why it needs all this energy. In contrast, our brains operate on about 20W, and however our brains learn things, they do it in a much smarter way than any AI. All that power and really, all AI can do is what we do only faster and with the ability to draw on a larger knowledge bank - it's not actually smarter than us (yet...) - and it still gets things wrong (and so do we...)
IDK.. I just have to wonder if we aren't barking up the wrong tree with AI in its current form (though god-knows what the right one is!) Nature, as usual though, came up with something all by itself that's pretty damned impressive.
Going back to the energy used (and other resources, most importantly, perhaps, water), it is somewhat irresponsible to dismiss this with a wave of the hand. Unless a viable alternative energy source (eg nuclear fusion) comes on line sooner rather than later, the planet simply cannot afford to carry on burning energy as it does. We will all pay a very heavy price otherwise.
I'm going to start backing up DOPUS installers.
I do not brush energy consumption aside, I was just trying to explain, that whenever energy "production" is getting cheaper, that energy will be sold and used. You probably won't see a decline in energy consumption any time soon.
I also did not say, that AI is "smarter" or more intelligent than a human person. It's a tool you can use to enhance on existing ideas and solutions, just like you make use of Directory Opus to handle 20 terabyte of files. You don't do that on the CLI anymore, you are using a proper tool for this, right?
The same applies now to "AI". If you want to create, let's say a performance benchmark for each coding language on each operating system. You can write the benchmarks by hand in some weeks, or just let the pattern engine create the code for you, it will take only some minutes and it won't make you dumb, because writing the same code 100 times does not require "intelligence", it is just exhausting to do (the same as trying to handle a zillion files on the command line).
What the outcome is, if even more energy is "created" and consumed, is another topic from my point of view. (We all know, energy cannot be created or used, you can only come up with new ways to transform energy.)
The CO2 emissions of Europe went down by 33% percent over the last 20-30 years, which does not mean we are using less energy, it's rather the opposite. Every kWh not spent on the fridge, because it is more energy efficient, will be used for the (new) air conditioning or for heating that sauna or charging the car or whatever kWh "eater" you have at home. It's not gonna revert unless energy get's more expensive, but it probably won't get more expensive, it will get cheaper, so even more will be used.
A good example of power efficiency getting ridiculously good and "wasted" at the same time, is with regular CPUs and GPUs. Some 15 years back, you had GPUs using 200-300W. Now after 15 years, the same graphics can be rendered with an integrated GPU in any CPU from Intel or AMD using maybe 20-30W, but there still are dedicated GPUs around eating 300W, 600W and more to render 3D games. Nobody settled with the GPU performance from 15 years ago.
Any technological advance will not lead to less power being used, as long as energy prices keep roughly the same or getting cheaper. Why does energy "production" get cheaper? Because of technological advance. You can't really stop this cycle I guess and there actually is no reason to, if the energy production is getting more and more friendly to the environment at the same time.
For whatever reason, quality of web search results has dropped dramatically for some years now, so users are basically forced to turn to the AI for answers.
I remember like 5 years ago, you could type almost any question into Google search and get useful results, and this used only a fraction of the power that LLMs use today.
I think the LLMs are a miracle technology, but companies are trying to shoehorn them everywhere these days.
Also, the normalized over-reliance on AI infrastructure controlled by large trans-national corporation entities can lead to dystopian scenarios, and IMO that's the main problem area of AI (not the technology itself).
But whether we like it or not, the way we use computers is changing once again, just like it changed when the internet became the norm. I too have a reflex to think AI is a needless crutch that will just "atrophy the muscles", but the same was thought of calculators once.
I still don't want AI in my Directory Opus though 
We live in interesting times.
Check this out: https://www.youtube.com/watch?v=t1eX_vvAlUc
I thought that the TPM chip was just cryptographic secrets, but apparently it does a hell of a lot more. According to Microsoft it can provide "remote attestation" (proving your PC's hardware/software state to remote parties) without your consent to anyone that sends a simple query. Pair that with A.I., crawling through your files and you've got something special.
Why? Well, you don't need a highly qualified 'tech-guy' to do a deep forensic dive. The A.I., is doing that 24/7 and your TPM chip is certifying it's your hardware's fingerprint. All you need is prompt the A.I., for what it is you want and the A.I., will fetch it for you.
Spooky stuff.
I don't want to speculate, but just looking at all this 1984 inspired horror facts... I can't help myself and think that framing someone in the future won't be difficult at all. These factoids don't even need to appear in the news... just sprinkle them in court... pile a bunch of circumstantial evidence... OK! Gonna stop myself now. This stuff is crazy.