Topic to complain about AI

Here you go, Rick and the esteemed band of complainers! 😉

The AI project, as you know, is the pinnacle of pure evil. It’s got everything—Microsoft secretly knowing about RISC OS, OpenAI lurking in the shadows, tons of code mysteriously “borrowed” from me! Truly awful stuff. And, of course, it’s here to change the world (or whatever other nefarious scheme you can dream up).

So, have at it! Complain to your heart’s content.

Stoke the boilers!

Everyone seems to love to scoff at AI, I’ve seen lots of comments like “oh it’s just glorified auto-completion”. But as far as I’m concerned, it’s like Star Trek fantasy technology made real.

It consistently blows my mind that I can just ask the computer “please make a song about Vladimir Putin being defeated. I said a song damn it, not an instrumental track, stop making instrumental tracks, also please include brass bands, SUPER brass, extreme brass”, in natural English, and it goes off and makes a whole album of it for me.
https://www.udio.com/playlists/aGd9s3JvkpS7f9Z9V1tvCn

For programming it’s been really helpful too. I’m not talking about getting it to write code for me, but rather, I use it as a supplement to reference manuals. It’s really nice being able to quickly ask stuff like for example “how do you get the fileno from a FILE*” and get an instant direct answer on the spot which is correct 99% of the time, rather than spending like 2 to 5 minutes searching through manuals or google results to find what I needed.
Sometimes, for more obscure stuff, it’s even been better than the official manuals or resources available online.

Here you go, Rick and the esteemed band of complainers!

Yeah… you might want to go back and read my post a little more carefully, as while I said that “AI” has a bit of a bad rap (let’s face it – it is mostly deserved), I was actually defending this project, pointing out that something specifically trained to know what RISC OS is might be useful.

The AI project, as you know, is the pinnacle of pure evil.

Gee, and I thought that wasn’t supposed to happen until next Tuesday.

It’s got everything—Microsoft secretly knowing about RISC OS

<looks at Windows 95 task bar> You’re about three decades late to that party.

Truly awful stuff. And, of course, it’s here to change the world (or whatever other nefarious scheme you can dream up).

Sounds just like religion. Does this new electronic god have a name? Or shall I just call it Tiamat II?

Stoke the boilers!

;)

Everyone seems to love to scoff at AI, I’ve seen lots of comments like “oh it’s just glorified auto-completion”. But as far as I’m concerned, it’s like Star Trek fantasy technology made real.

Patrick, absolutely! I’m just being a bit sarcastic with “the usuals”, they’re all good folks, just fond of turning most threads here into “grumpy mode”. Thus, my nickname for them (inspired by the famous band Foreigners): “The Complainers” 😆. But the reality remains that AI is a powerful tool—just that, a tool.

They’re not wrong, though, in venting about the “AI scam” hype around the world. It’s less about AI as a tool and more about people (often with fancy degrees) finding ways to make big money with overblown promises (like we’ve seen with quantum computing, string theory, and, unfortunately, even some cancer research areas etc.).

So yes, maybe 60% of the AI buzz out there is just hype. I’ve been developing AI since I was young, and I’ve seen my share of it. But that doesn’t take away from the genuinely useful applications of AI. Don’t let “The Complainers” dissuade you; hopefully, this tool will be a real help to those navigating RISC OS and those looking for more knowledgeable content. However please do not be passive, it takes a lot of work to get an LLM to provide useful answers in most cases, so join me in making the model better. The Complainers won’t help, hence I have no interest in their shenanigans.

Yeah… you might want to go back and read my post a little more carefully, as while I said that “AI” has a bit of a bad rap (let’s face it – it is mostly deserved), I was actually defending this project, pointing out that something specifically trained to know what RISC OS is might be useful.

I did read all the comments actually, which is why I ask to move the complains in Aldershot, I have managed to ignore it, which the comments on other places here trigger a notification in my software, in case a user needs help getting things working. The various opinions aren’t anything useful to that end.

I thank you for your review, but there was no mistake in calling the topic Generative AI, because that is what the tool does.

Gee, and I thought that wasn’t supposed to happen until next Tuesday.

LOL sorry! The Borgs do not wait, any resistance is futile! XD

You’re about three decades late to that party.

Indeed very suspicious, we need to get revenge! ;)

Sounds just like religion. Does this new electronic god have a name? Or shall I just call it Tiamat II?

Oh believe me, only a bunch of fools take it as a religion, every one else around it is just trying to “milk easy money” my friend :)

It has it’s uses. They may or may not be serious, I don’t know. But it writes bedtime stories for my kids, starring my kids (and their dogs) about three times a week!

Currently, thanks to AI, I’m the best dad in the world ever. They have a very low bar, but I’ll settle for that.

every one else around it is just trying to “milk easy money”

That sounds like a description of a religion to me.

Currently, thanks to AI, I’m the best dad in the world ever. They have a very low bar, but I’ll settle for that.

You can be replaced you know – quite easily, by the look of things ;)

I’ll paste this here to avoid polluting the other thread. I asked ChatGPT.

You may want to sit down.

And put that mug of tea down.

And those Ginger Nuts.

How do I use Wimp_PollIdle?

Wimp_PollIdle is a function used in programming for handling idle events in the WIMP (Window, Icon, Mouse, Pointer) environment, particularly in RISC OS applications. Here’s a basic overview of how to use it:

* Initialize Your Application: Make sure your application is set up properly, and you have created the necessary event loop.

* Set Up Idle Handling: You typically call Wimp_PollIdle within your main event loop. This function checks for idle events and allows your application to process them.

* Poll for Events: In your loop, call Wimp_PollIdle after handling other events. This will allow the application to respond to idle events when there are no other user inputs.

* Handle Idle Tasks: Implement the actions you want to perform during idle times within your application. This could include tasks like updating the UI, processing data, or performing background operations.

Example Loop:

while (running) {

    Wimp_Poll(); // Handle other events

    Wimp_PollIdle(); // Handle idle processing

    // Any other code you want to run

}

Considerations:

* Ensure that your idle tasks are efficient to avoid blocking the user interface.

* Use Wimp_PollIdle judiciously to keep your application responsive.

* This should give you a solid starting point for using Wimp_PollIdle in your applications!

But it writes bedtime stories for my kids, starring my kids

Wouldn’t it be more, I don’t know, “personal” maybe, to come up with your own stories?
Maybe even incorporate things that have happened to them recently (new achievement unlocked, you opened the jam jar by yourself!), that sort of thing?
Of course that’s not to say I wouldn’t cheat too. Hey, sweetie, you’re five now. Let’s start with some longer short stories. Here are a few written by a guy about a hundred years ago. His name was Lovecraft… 😉

Yes, so, the fact that both use OpenAI LLM, doesn’t mean that ChatGPT (standard) actually can be used instead of the RISC OS Assistant.

I assume that at some point (in the future), by us using the RO Assistant, OpenAI will integrate the content in their LLM, but when this will happen I have no idea. I just assume they use data from everyone and everything, so eventually also from this.

Just for comparison, Google Gemini was utterly hopeless, even when provided with the answer in an earlier prompt in the same thread.

Google Notebook however, given a PDF of Volume 3 of the PRMs (1) and a link to this site (2), answered:

Wimp_PollIdle Parameters and Usage
The Wimp_PollIdle SWI (1) has the following parameters:
● Entry:
○ R0: Poll Mask (1)(2)
○ R1: Pointer to 256 byte block (1)(2)
○ R2: Earliest monotonic time for return with null reason (1)(2)
○ R3: Pointer to pollword in RMA (if R0 bit22 is set) (1)

● Exit:
○ R0: Reason Code (1)
○ R1: Pointer to block of data (1)
○ R2: Senders task handle (1)

Wimp_PollIdle functions similarly to Wimp_Poll (1)(2), but it incorporates a time parameter (R2) that allows the task to “sleep” until the specified time has passed or a non-null event occurs (1)(2). This means that Wimp_PollIdle won’t return a Null_Reason_Code before the time given in R2, but delivery of other events is not affected (2).

A programmer would choose Wimp_PollIdle over Wimp_Poll when their application needs to perform actions at scheduled times (2). This is because Wimp_PollIdle provides an efficient way for a task to sleep until the next action is due (2). This allows other tasks to have more processing time, improving system efficiency (1).

For example, if a task needs to perform an action every 100 centiseconds, it can use Wimp_PollIdle to wait until the next 100 centisecond mark before waking up and performing the action (1). This avoids the task constantly waking up and checking the time, wasting processing time.
However, it is important to note that there is no guarantee that the task will regain control at precisely the time specified (2). The system could be busy servicing another task, or a single-tasking program could have started (2).

(There were various citations to the sources, I’ve limited them to the site)

This is a very good answer given the source materials. Note that it correctly uses “monotonic time” rather than merely “system”, and understands that Wimp_Poll returns constantly (if you’ve not masked nulls). It is noticeable that it spends some time ‘thinking’ about the question, and of course it knows nothing out of the box.

It also provided this summary:

Summary
This source is a programmer’s reference manual for the RISC OS 3 operating system, produced by Acorn Computers in 1989. It provides a detailed overview of the RISC OS kernel, filing systems and Window Manager (Wimp), covering topics such as memory management, file operations, window creation and event handling. The manual also includes technical details on how to write your own filing systems and Wimp applications. The manual goes into depth on how programmers should format and design their applications to be consistent with the RISC OS environment.

And suggested questions such as:

How does RISC OS handle multi-tasking and communication between applications?
What are the design principles and conventions for developing WIMP applications in RISC OS?

This is also the AI that can generate a conversational audio podcast between two deeply American hosts who discuss any subject that can be asked of the sources. It’s really rather impressive, and I’d recommend you try it out

monotonic time

That’s impressive.

This source is a programmer’s reference manual for the RISC OS 3 operating system, produced by Acorn Computers in 1989

That’s impressive, in a different way.

Indeed, and it’s able to do this without training – it just reads the sources you provide.

I imagine it’s heated the planet by 0.1°C in the process, but hey, you don’t have a dog-that-can-read and bark yourself.

RISC OS 3 in 1989?

Remember AI understands NOTHING. It is purely a statistical word guessing algorithm, with no ability to determine if its output is correct.

Such a system is worse than useless for technical documentation.

Druck,
honestly I don’t understand why you are soooooo defensive man. Let it happen, trust me, it will be quicker than having folks like you fighting back, you are just slowing down a wave of madness that needs to pass quickly instead:

https://www.euronews.com/culture/2024/11/01/thousands-go-to-fake-ai-invented-dublin-halloween-parade

Read it, it’s hillarious! An AI invented a news, other just trusted it, thousands showed up, we have the pictures of the fools basically (sorry for the language).

They now have learned their lesson and if not, they’ll do again and get the same results. At work it has to be the same, believe me, when companies will be crippled down by the bugs, our job will be worth billions and they’ll offer us the world to help them getting out of their misery.

From the article linked:

While this has been largely received as a funny incident for the many Irish Halloween fans duped into believing in the fictitious parade, it’s an alarming display for how powerful misinformation can be.

Druck, honestly I don’t understand why you are soooooo defensive man.

Right, say I’ve forgotten how something in RISC OS works, so I google for it. I could look at the AI generated nonsense at the top, and try that finding invariably it was wrong, or I could take a few extra seconds to look what websites are in the first couple of results, and go straight to the authoritative and correct information from RISC OS Open. If I needed more help, there are other hits from people I recognise from this forum.

There really is no way I would consider trusting a statistical mash up of random crap from the internet.

Right, say I’ve forgotten how something in RISC OS works, so I google for it. I could look at the AI generated nonsense at the top, and try that finding invariably it was wrong, or I could take a few extra seconds to look what websites are in the first couple of results, and go straight to the authoritative and correct information from RISC OS Open. If I needed more help, there are other hits from people I recognise from this forum.

Ok, I just did this test and asked Google how I can copy files on RISC OS. Here is the response:

1) The “authoritative” source (for Google) is not the ROOL site; it’s riscos.com. <—- Yup, one man vs. all of us managed to have better documentation and has a full page dedicated to copying, moving, and renaming files using the Desktop, so first hallucination for Druck’s neurons here.

2) This was followed by a list of results that did NOT answer the question (but it’s ok; riscos.com DID answer the question).

3) ROOL was far down in the list of results, with forum pages—not authoritative documentation. And given the behaviour of quite a few on here, I let you decide the actual “look” of it XD

Let’s do another test. Let’s search for a modern and really small footprint OS for ARM:

RISC OS appears on the first page results! ( XD – ridiculous obviously ) and that is because between the so-called authoritative source and some insane articles it has/had quite a few folks posting content about RISC OS being modern (and we all know RISC OS is as modern as the BBC Micro MOS ported to ARM ;) ), and given that Google’s indexing is even dumber than an LLM, well, you get the picture.

So, AI vs. Druck—I’d say both are hallucinating here, and this is the problem. There isn’t a right vs. wrong. RISC OS is niche and old, and it requires US to keep it alive and being indexed correctly.

There are many reasons why we SHOULD work on preparing AI training data for RISC OS, because it will be the future, and when that becomes the norm, Rick up here has already shown us what will happen (given no one cares about organizing RISC OS knowledge for AI).

You’re fighting a war that you (and I would say: we all) have already lost. Everyone is entitled to their own opinions, always! So, I won’t try to change yours, but your constant and repetitive comments on the same stuff won’t change what’s happening out there.

It’s a weird world when we think about it. Our nature drives us to preserve energy (the major reason behind the brain’s development of memory and use of experience), and yet we live in a world that wants to learn from the ground up again and replace us with machines to avoid paying salaries and make more profits. However, these very same salaries are what give these folks their profits lol—and here we are discussing AI hallucinating… I mean, are we sure we’re focusing on the right subject?

one man vs. all of us managed to have better documentation

Wasn’t that an automated translation of the official Acorn documentation? In other words, the text of the PRMs/user guides?

You’re fighting a war that you (and I would say: we all) have already lost.

AI is not helping, but really the major enshittification of reality has been the creeping tentacles of advertising. It’s ever harder to avoid that crap.

but your constant and repetitive comments on the same stuff won’t change what’s happening out there.

Don Quixote tried, and failed, and now the place is crawling with wind turbines.
Can’t help but think that maybe he was on to something, a visionary way ahead of his time.

what give these folks their profits lol

No, modern capitalism is about getting rid of excess personnel to increase profits because those who would buy the product are somebody else’s employees.
Yes, it’s a circular argument, but there’s plenty of evidence that few look beyond the next quarter earnings…

Buy mining shares. More copper needed. (This is not advice :-))

Wasn’t that an automated translation of the official Acorn documentation? In other words, the text of the PRMs/user guides?

AFAIK Yes + added human written articles focused on specific and common users questions. Still one man vs the rest of us in this case and certainly (for Google) more authoritative.

It’s ever harder to avoid that crap.

I agree on your definition, while I disagree on this statement, more and more people are blocking ads. I had ads on my blog for decades and the profits have been going down since 2016 IIRC. I don’t fear to share, given I know people wants more and more for absolute free and without any desire to help, so there are issues both sides if you ask me. Selfentitlement is also a big social problem at the moment.

Don Quixote tried, and failed, and now the place is crawling with wind turbines. Can’t help but think that maybe he was on to something, a visionary way ahead of his time.

XD He certainly IS representative of a portion of our society as well, indeed. Was him onto something? Maybe.

No, modern capitalism is about getting rid of excess personnel to increase profits because those who would buy the product are somebody else’s employees. Yes, it’s a circular argument, but there’s plenty of evidence that few look beyond the next quarter earnings…

It’s interesting that you’ve started a sentence that agrees with my point with a “No,”. I though in English you would have written something on the lines of: Indeed, modern capitalism is about getting rid of excess personnel to increase…

Maybe something got lost in the translation.

But yes, no one is thinking about medium terms (let alone long term), it’s all in the here and now. Soon tech jobs will open again, because the amount of problems is growing at a faster rate than ever before.

[edit]
More examples of how the world is changing really fast: https://x.com/Rainmaker1973/status/1853452406597538244

This is how is it possible in this day and age to create very professional ads using just and iPhone etc. stuff that would have had a much bigger cost not long ago. So, it’s not just AI causing problems with jobs (probably AI is not even the biggest threat at the moment), it’s the whole technological progress.

Beside: Guys we were the kids that wanted GUIs over terminal, some of us worked hard to make the Internet a world wide thing. Fighting against AI, sounds a bit like the old folks that kept their teletype. Did GUIs change everything? YES, did they solve all issues? NO. Did the next revolution after the GUIs change everything (the WEB Apps) YES, did it solve all issues? NO. Did Cloud Computing change everything? YES. Did it solve everything? NO. Did Mobile Computing change everything? YES, did it solve all tech problems? NO. It’ll be the same for AI, we are just in the hype-bubble-time right now, when this is over it’ll be just another tool.
[/edit]

and certainly (for Google) more authoritative.

Don’t fall into the trap. Google doesn’t decide what is and isn’t authoritative. It simply tracks what is popular (by how often the link is followed) and popular rises to the top, which makes it something of a self fulfilling prophecy given the number of people that’ll just follow the top link…

…or, to put it an alternative way, Google’s idea of authoritative is to point at a branch of the OS that died out many years ago and is still sort of hanging around like one of the undead. Once upon a time, it probably got a lot of traffic. Maybe still does as there’s some stuff there that’s easier to find than reading the wiki here.

more and more people are blocking ads.

Correction: More and more people that know how to are blocking adverts. The browser makers are making it increasingly hard to do so (massive conflict of interest given that their companies tend to make a lot of coin from advert revenue), and adverts in apps are more and more insidious. Where it was a banner or pop-up, it’s now often a one-two minute video of a dumb game followed by a playable demo that also lasts about a minute or so.
Or the much shorter advert for Tenu THAT ALWAYS BEGINS WITH THE VOLUME AT MAXIMUM.

Setting up things like DNS blocking or PiHole are not skills that the average user has. So arguably while more and more techie users are blocking adverts (with varying degrees of success), far more people are being exposed to even more in-your-face bullshit.

Oh, and Prime video and Netflix alike are dropping in adverts into the programming, although to date I’ve only seen one advert. Maybe there’s a certain amount of programming one needs to watch first? 🤷🏻‍♀️

I had ads on my blog for decades

I never have and never will. I have too much self respect to force people who willingly choose to read the crap I write to have to run unknown scripts and content from random third parties that I have no control over.
If I like something, I’ll talk about it. If I don’t, I’ll talk about it more.

Selfentitlement is also a big social problem at the moment.

Is this the point where we oldies mumble “millennials” in the most disparaging tone possible? ;)

It’s… interesting at work. Some of the younger workers have no clue or concern about other people. It’s pretty normal to hold the door open if there’s somebody behind you, but far too often to just be one person, a younger worker will breeze through and ignore anybody and anything around.
There’s just something wrong with people’s attitudes, or just really really lousy parenting.

Maybe something got lost in the translation.

I took it rather more literally in that you were suggesting that by employing these people, it gives them the money to buy the stuff; so my “no” was to correct you to suggest that they can lay off their workers because other companies workers will buy the stuff.

Of course, you’re not supposed to ask what happens when nobody has a job any more and nobody is buying stuff…

Soon tech jobs will open again, because the amount of problems is growing at a faster rate than ever before.

Well, they’ll need people to turn the rubbish that AI generates into something useful. ;)

This is how is it possible in this day and age to create very professional ads using just and iPhone etc.

The videos on my YouTube are never made with much attention to “production values” (my budget is precisely zero so…) however it’s now quite possible to record video, edit with numerous effects, chromakey myself onto whatever background I choose, add in on-screen text, blah blah blah, just using a free app. All at 720p or 1080p (could do 4K but I don’t have enough free space on my phone for that).
What a generic Xiaomi device can do now, compared to just a few years back…

And that’s without anything that resembles AI; though in my most recent video (the spooky mindscrew), I did use an AI art app to create the backgrounds, and various comically wrong diagrams. ;)

it’s the whole technological progress.

True, but there’s also the factor of how much utter rubbish this generates. If you look at what trends, you’ll start to see a pattern of the same creators being repeated quite frequently – often “the sidemen” and “Mr Beast” (I have no idea who either of those are, it’s just something I notice given my YouTube player starts at the “trending” screen by default). That’s a drop in the ocean compared to how much content gets uploaded that few people ever watch, but that stuff, it got created, it got uploaded, it’s floating around like a turd in the bowl just like this message (and the other 13,800 that I’ve written). Technological progress is drowning in dross.

Guys we were the kids that wanted GUIs over terminal

I was the kid that wanted a computer… what it could do came afterwards.
Thankfully I talked mom into getting a second hand Beeb, and a few years later an A3000.

Fighting against AI, sounds a bit like the old folks that kept their teletype. Did GUIs change everything? YES, did they solve all issues? NO.

Ah, but there’s a big logical flaw in your argument. The GUI was an attempt to make a more user-friendly interface.
It was fairly commonplace to have a text-based windowing system back in the day, think TurboC or WordPerfect. So it wasn’t actually that much of a leap to transition to a full graphical version once VGA became common and had enough on-screen pixels and colours to make it actually work.

AI, on the other hand, is this bizarre fantasy where far too many think it’ll ingest the entirety of Reddit and spit out the cure for cancer.

Personally I think the whole AI thing will eventually blow over, and then real programmers can get on with making proper artificial intelligence aids without all that crap being spouted about the massive benefits of LLMs.

Alas, I think we first have to endure the period of companies attempting to stuff them into anything and everything, like flinging the brown stuff at the wall to see if it sticks.

Some such things can be useful, like software assisted photography to try to recognise what the photo is and set up the camera to get the best photo. Funny thing is, my phone calls it AI. My little digital camera can do the same thing but it predates the AI craze so it simply calls it “Assisted mode” (select that, push the shutter halfway, let it decide the best settings…). It can also detect faces, another thing called “AI” these days.

when this is over it’ll be just another tool.

Yes, I fully agree. But I also agree that one should feel free to argue down the hype because, well, just read…

https://www.theregister.com/2024/10/29/softbank_super_ai/