Is RISC OS still a participation sport

Artificial it is, Intelligence it falls short of.

Jury is out. Meanwhile we keep moving the goal posts.
After Chess succumbed people said well that was not intelliegence, but there is no way a machine can win at Go. Well that did not take long.
So obviously you don’t need intelligence to play Go.
I wonder how we are getting on with the Turing test.
When or if that is passed will we say that was not a test of intelligence either?

Well… Being good at chess and go is the ability to calculate many many moves ahead and to do that for the opponent to decide upon the best move to make. In essence, it is pattern matching that doesn’t require “intelligence” so much as the ability to make many calculations at the same time. The sort of thing machines are good at.

It’s been said that the most important thing about Go is not whether or not you win, but how you play.

My personal belief is that machines are good when it comes to making calculations. It may be able to design a better wing for a plane because it has all the maths for how air passes around the surface. So it can just rough out a few hundred randomly generated “wing shaped” objects and run a rough simulation. Those with good results can pass to a more complete simulation, and whittled down that way.

But is the AI ever going to be possible to point at the wing and say “this part makes the difference”, or will it simply say “this design resulted in the best score”?

Where humans excel, however, is in abstract thought. Conceiving things that don’t exist.

Anna (my cat the cat that owns this human) may hear a noise, consider the possibility that it’s a mouse, home in, line up, wait for it, wait for it, wait for it, GO!GO!GO!GO!GO!

We, on the other hand, may observe a mouse. Watch it, watch where it goes and how it moves, then think about what sort of object might catch it. Then think about how to build that. Get some hardware (nuts, bolts, wood, etc) and do some work to turn the hardware into the pieces to build the trap. Which can then be built, put in place, and observed to see if it works.

We are so good at dealing with abstract concepts that we’ve invented entire religions to try to explain our presence. Because we want to understand “who am I?” and “why am I here?”. The sort of existential questions that it’s unlikely that any other lifeform on the planet contemplates, and a machine never could.

It’s been said that the most important thing about Go is not whether or not you win, but how you play

Agreed – buts thats true of all games. What I like about Go is the emphasis on the sound of the stone hitting the board amplified by the integral cavity designed to amplify it.

To further this, since we’re in Aldershot, it’s coming to Spring time and that’s when Anna gets a bit clingy, rather than her usual “you may stroke me once and only once, human”.

Why? She’s female, I’m male, hormones are involved, and that’s as far as it goes.
She has no concept that she’s a cat. And to her, I’m an odd shaped creature that isn’t a predator but provides food plucked from out of thin air (actually a pocket, but…).

I know I’m human, I know she’s cat, I know sex between us wouldn’t work for very many reasons most of which too squicky to get into.

She….. doesn’t.

So I might sit in my deck chair looking at the sky wondering why I’m here.
She’ll just amble over and be like “feed me”, even if I’ve just done so.

Could an AI be a cat? Perhaps. (but don’t tell Anna I said that or she’ll be offended and scratch my eyes out)

Could an AI be a human? Nope.

The biggest issue I see with this is that the terms are really just marketing hype. Smart doesn’t mean ‘clever’ in some way – it only means internet-connected, extracting data about your use and limiting the value of said smartness to ‘as long as we support, maintain and stay in business’ while giving some useful information, access or other goodies back to you – whether you want them or not.

Likewise, AI mostly isn’t intelligence, it’s training to look for patterns and inferences. In the earlier days of automation, robot car body sprayers were ‘taught’ by a skilled sprayer doing the job and that motion, duration, angle, etc. fed back into the robot, which then repeated the action over and over. A lot of AI seems to be the processing version of the same thing. Is there something on this X-ray that looks out of place compared to the images it was trained on? Given a section of text, what keywords and phrases can it find to get the gist of what the person is asking and how much data does it have access to that will enable it to make a meaningful response? It would be worth feeding it some posts by amanfrommars1 on the El Reg comments to see what it could make from them!

It will get better, but I’m not seeing any intelligence yet.

It would be worth feeding it some posts by amanfrommars1

👍👍👍

It’s hard enough for us meatsacks to work out what the hell he’s on about. I’d imagine a machine would have sparking control panels followed by clouds of smoke.

Smart doesn’t mean ‘clever’ in some way

This: smart is used in the context of “not dumb”, where “dumb” is a ridiculously low bar to allow just about anything “with internet access” to call itself smart.

A washing machine that weighs the load and works out an appropriate cycle is not “smart”. A smart washing machine would spit out the red t-shirt you put in with all that white stuff, and understand that wool and cotton get washed differently without the user having to say “I want an IWS approved wash”.
We’re a long way from that.

Where humans excel, however, is in abstract thought.

Says the human. How do you define “abstract thought”? Because if it’s a mean to an end, then it’s not abstract, just complex.

Birds show plenty of problem-solving ability. Chimps have a greater sense of number than humans. Humans, meanwhile, have absolutely no sense of probability, chance, odds or percentages, despite all the lofty talk of “abstract thought”.

We motile living things evolved the sense of number primarily to avoid being eaten by tigers – you see tigers over there; you look away; if you look back and there are fewer tigers, you could be lunch.

But we have not evolved any equivalent understanding of probability. Instead we have a set of poor heuristics that not only get the wrong answer – in common cases they get the answer as wrong as it is possible to be (Linda Question; Monty Hall).

I suspect that any species that thought a “third-pounder” was smaller than a “quarter-pounder” should hold back on the “we’re good at thinkin innit?”.

Intelligence Sentience Spanner

Pop!

How do you define “abstract thought”?

The ability to understand and think about things that we have not observed, and potentially cannot observe.
Examples? From flying on Mars (getting to Mars!) to most theoretical science.

Maths, also, rapidly moves from the “count these things” into abstract concepts.

I suspect that any species that thought a “third-pounder” was smaller than a “quarter-pounder”

Please don’t conflate innate mathematical and abstract problem solving abilities with shitty education.
Because fractions are themselves a bit abstract, people need to be guided to understand that if the bottom number is bigger, you get less.
That’s what teachers are for.

I would love to make a glib comment about Leftpondians, but then I recall my school days where the popular kids were the dumb disruptive ones, and the “clods” would be routinely wedgied.

The ability to understand and think about things that we have not observed

Has been seen in various animals. Object permanence, inference and prediction are not exclusively human traits.

Regardless, this is already settled Law – all vertebrates and some invertebrates are sentient.

Is there something on this X-ray that looks out of place compared to the images it was trained on?

If so, refer to human for full checks

Has been seen in various animals. Object permanence, inference and prediction are not exclusively human traits.

Cat named “Minnie”¹ who will observe a spot of light moving on the floor, glance at the sunlight coming in, swing round at the correct angle and then stare at me (or the wife) with a look that asks the question “does that amuse your small mind?”
Dumb animals? Pah!

¹ Mistakes you make, she was small when she turned up. A few years of excellent food and I refer to her as Pansy Porker more often than Minnie

The ability to understand and think about things that we have not observed
Has been seen in various animals.

I have kept chickens for many years. Just two or three at a time. Few enough to get to know them personally and viceversa. They are protected in law against physical cruelty but not from stereotypical abuse. They are birdbrained but when it comes to outwitting them I have only just got the edge, and this only because I have experience of 20 odd generations of their kind. They are sentient and intelligent.

Interesting comparing brain sizes – what does a large human brain do?

How does a AI computer compare with a chicken :-)

what does a large human brain do?

Daydream.

This would be a good time to test to see if it could translate arm code into C then we may be able to save RISC OS with the finite number of developers we have. Anyone up for it?

How does a AI computer compare with a chicken :-)

Here’s the problem, or rather was before the birds were confined to the run owing to bird flu,
How to get them in at night. In general they will go to their run when they deem it is close to bedtime. This varies according to the time of year and the weather. If the weather is gloomy they will go in a little earlier than on a bright evening. The problem comes when I want to go out for the evening and need to put them to bed early. They can’t be left out because of fox hazard. Chuck a handful of corn in the run and one or two birds might go in, but no guarantee. It is not possible to chase them in. The run is the last place they will go, and they are past masters of the nutmeg.
AI does not seem to offer a solution here.

Have you tried a bit of mince meat?

They’ll jump up and take it from your fingers :-)

This would be a good time to test to see if it could translate arm code into C then we may be able to save RISC OS with the finite number of developers we have. Anyone up for it?

Q. Can you convert arm assembly language into C code ?

A. Yes, it’s possible to convert assembly language into C code. The process involves understanding the assembly instructions and then mapping them to equivalent C statements. However, this can be a complex and time-consuming task, as assembly code is often highly optimized for a specific architecture and may not have a straightforward mapping to C. Additionally, the behavior of the resulting C code may not be exactly the same as the original assembly code, due to differences in the way that C and assembly handle certain low-level operations.

AI does not seem to offer a solution here.

NI does. Ladies tend to fare better when wearing skirts, particularly long ones. Nutmeg techniques don’t work when trying to pass through a sheet of material.
Farmers tend to wear long aprons for the task, some people use sheets of board or netting.

There is the alternate: select the brighter ones that like hopping up into low bushes and trees. An estate worker/gamekeeper recommended that.

ChatGPT believes that it can create BBC Basic code (e.g. “create a programme to plot a mandelbrot picture on a computer screen”). However, the results contain lots of rookie errors for BBC Basic – like confusing some BBC Basic statements and functions with their VB equivalents. But most of these can be corrected just by asking ChatGPT to fix them – i.e. just by pointing out the nature of the errors.

If ChatGBT was able to learn from the corrections that ‘weighted’ outside users (rather than ‘trainers’) were making, then this might become a very useful tool very quickly. I guess the hard bit is weighting the clever users (e.g. Jeffrey or Gerth etc.) much more that us mere ‘normals’.

Note: Python code creation seems to be pretty good using ChatGPT – but given that Python is so well structured and so widely used then this is not surprising.

This would be a good time to test to see if it could translate arm code into C then we may be able to save RISC OS with the finite number of developers we have. Anyone up for it?

Some years ago, before I retired, I was asked to examine the possibility of getting some Visual Basic translated by machine into VB .Net. Having examined what was on offer, I decided that the result would be very difficult to comprehend and therefore unmaintainable. And that was for two languages that, on the face of it, have some similarity.

I’m sure the same difficulty would exist when trying to translate ARM assembly language into C. Certainly the same question must be asked: will the result be fit for humans to modify and maintain?

stare at me (or the wife) with a look

Theory of mind. Often their theory is “useful idiot”.

save RISC OS

Are the faeries coming to take it away?

RISC OS is an operating system for running RISC OS programs. RISC OS programs are either Basic or ARM machine code. ARM machine code is trivial to run whatever CPU you happen to have this week – some of us have been running RISC OS every day for 20 years without an ARM.

You don’t need to ‘convert to C’ to ‘save’ RISC OS.

What you need is some basic competence at operating system maintenance. Changing the subject completely, when is RO5 OS_SetVarVal going to become compatible with the 2004 RISC OS API?

<looks at all and sundry like Minnie looks at Steve>

Seriously???

RISC OS programs are either Basic or ARM machine code.
OK for things compiled from C or such you can say “machine code” – but how about Python, PHP, Lua, …?

ARM machine code is trivial to run whatever CPU you happen to have this week
Wow, thanks to explain why I needed so many software updates for things to run on StrongARM and why I still need Aemulor to run publisher after reading this statement.

You don’t need to ‘convert to C’ to ‘save’ RISC OS.
I strongly dis-agree. Sure you can do it all in assembler code – but that is a lot more work to maintain and if needed update to work with a new ARM CPU variant.
Saving RISC OS does not just mean have some old hardware that can run it – or emulate it – but make it run on current hardware natively since if that is not part of the package it is “preserve” only.
If you want to be able to maintain and enhance something as big as RISC OS using Assembler is not a good idea since it is a lot more work to code and a good compiler even does a lot of checks for you that can save lots of debugging frustration