AI railway

[FatherMcD]

The problems start when people take it as gospel, start to ignore the errors, and stop listening to experience

When I point out a clear mistake made by the AI I use for research, it generally compliments me for catching it, but in general, doesn't actually acknowledge it's mistake. Almost like it doesn't actually have any sense of self-awareness.

 

[PhilP]

When I point out a clear mistake made by the AI I use for research, it generally compliments me for catching it, but in general, doesn't actually acknowledge it's mistake. Almost like it doesn't actually have any sense of self-awareness.

Thank god!

Imagine how sarcastic AI will be when it becomes self aware!

PhilP.

 

[Rhinochugger]

When I point out a clear mistake made by the AI I use for research, it generally compliments me for catching it, but in general, doesn't actually acknowledge it's mistake. Almost like it doesn't actually have any sense of self-awareness.

Yes, remember it is 'artificial'. It can be useful for those that embrace it but, as others have said, it needs checking, and I don't see how that will change.

We need to remember that the difference between man and machine is that human beings can make a judgement - a machine can only provide a response based on previous or programmed data.

I remember a TV programme that Jeremy Clarkson did travelling America, and in one episode he was introduced to a robot which could 'answer any question'.

Clarkson, being Clarkson, asked his first question - 'What do you think of the Dooby Brothers?'

Result: Clarkson 1 - Robot 0

(and no, the robot couldn't even respond that it hadn't heard of the Dooby Brothers.)

 

[Madman]

All good points on AI. As someone once said, "Know your enemy".

 

[Djjerme]

AI can be useful when it’s treated as a tool, just like anything else in the hobby.


I actually wrote an article for Garden Trains Annual about using AI for modeling purposes, and my experience has been that it works best when it supports the skills you already have rather than replacing them. As a software developer by day, I’ve used different LLMs for code debugging, testing, and handling mundane tasks like fleshing out API endpoints or basic class structures. But you absolutely still have to proofread and sanity-check anything it produces — as the “photo” mockup in the original post demonstrates pretty clearly.

For modeling, I’ve found ChatGPT helpful for deep research, organizing notes and ideas, and even turning handwritten notes into something searchable so I can revisit concepts, measurements, and design decisions later.

One project I’m working on right now combines 3D printing and AI to design a modular building system for structures. The initial research is still the same as it has always been: looking at real-life examples, photos, plans, measurements, sketches, and historical references. From there, I used ChatGPT to help work through available space measurements and design cardboard mockups. That let me physically tweak the structure sizing before committing to anything more permanent.

Once the sizing was close, I used Codex to help refine panel dimensions for 3D-printable test pieces: checking clearances, interlocking panel methods, material thickness, and repeatability. Since I’m a code monkey, I used OpenSCAD for the 3D modeling side. That gave me more precise control than something like TinkerCAD and also made it easier to refine the design consistently in code.

But that’s the key: it’s all just tools.

It’s not taking away anyone’s job, creativity, or modeling skills. It just changes which skills get used and when. Could I have scratch-built something quicker? Probably. I’m already three-plus months into the project and still in the prototype phase. But it has given me a lot of insight into where AI can genuinely help, especially with planning, iteration, documentation, and working through design options before cutting or printing parts.

One thing I mention in the article — and something that shows up in examples like the first post — is that AI is very good at trying to make everything work. Sometimes too good. It can be overly encouraging, and it rarely stops you and says, “Maybe this is a bad idea.” If you truly believe you can fly, AI may help you design the wings instead of pulling you back from the cliff edge.

So the important part is to step back from the result, look at it with your own experience, and ask whether it actually makes sense.

Or, as Dr. Ian Malcolm put it: “Your scientists were so preoccupied with whether they could, they didn’t stop to think if they should.”

 

[FatherMcD]

If you truly believe you can fly, AI may help you design the wings instead of pulling you back from the cliff edge.

AI has even been known to suggest that you try! Any tool can be dangerous if you don't know how to use it or get sloppy.

 

[GAP]

The problems start when people take it as gospel, start to ignore the errors, and stop listening to experience

Bit like listening to "Influencers" on social media and swallowing what we see on the telly. We really are a gullible breed aren't we?

 

[GAP]

AI can be useful when it’s treated as a tool, just like anything else in the hobby.


I actually wrote an article for Garden Trains Annual about using AI for modeling purposes, and my experience has been that it works best when it supports the skills you already have rather than replacing them. As a software developer by day, I’ve used different LLMs for code debugging, testing, and handling mundane tasks like fleshing out API endpoints or basic class structures. But you absolutely still have to proofread and sanity-check anything it produces — as the “photo” mockup in the original post demonstrates pretty clearly.

For modeling, I’ve found ChatGPT helpful for deep research, organizing notes and ideas, and even turning handwritten notes into something searchable so I can revisit concepts, measurements, and design decisions later.

One project I’m working on right now combines 3D printing and AI to design a modular building system for structures. The initial research is still the same as it has always been: looking at real-life examples, photos, plans, measurements, sketches, and historical references. From there, I used ChatGPT to help work through available space measurements and design cardboard mockups. That let me physically tweak the structure sizing before committing to anything more permanent.

Once the sizing was close, I used Codex to help refine panel dimensions for 3D-printable test pieces: checking clearances, interlocking panel methods, material thickness, and repeatability. Since I’m a code monkey, I used OpenSCAD for the 3D modeling side. That gave me more precise control than something like TinkerCAD and also made it easier to refine the design consistently in code.

But that’s the key: it’s all just tools.

It’s not taking away anyone’s job, creativity, or modeling skills. It just changes which skills get used and when. Could I have scratch-built something quicker? Probably. I’m already three-plus months into the project and still in the prototype phase. But it has given me a lot of insight into where AI can genuinely help, especially with planning, iteration, documentation, and working through design options before cutting or printing parts.

One thing I mention in the article — and something that shows up in examples like the first post — is that AI is very good at trying to make everything work. Sometimes too good. It can be overly encouraging, and it rarely stops you and says, “Maybe this is a bad idea.” If you truly believe you can fly, AI may help you design the wings instead of pulling you back from the cliff edge.

So the important part is to step back from the result, look at it with your own experience, and ask whether it actually makes sense.

Or, as Dr. Ian Malcolm put it: “Your scientists were so preoccupied with whether they could, they didn’t stop to think if they should.”

I agree; not being a "Code Monkey" I used it to write a program from my description of what I wanted the finished product to do and then tweaked it till it did exactly that plus a few new extras.
It is a tool in my case a DIY program writing one, also used it to write microcontroller programs.
I also used it to get a rough design of a few accessories that I would like to build. The useful bit of that was that it scoured the net to find something that would work saved me umpteen hours of scrolling on google. That is one example of how it can be used to help.
Its not going to take over the world unless we let it; it cannot "think" for itself like we can (that last bit is debatable in some cases).

 

[Madman]

Or, as Dr. Ian Malcolm put it: “Your scientists were so preoccupied with whether they could, they didn’t stop to think if they should.”

And therein lies reality, in my opinion. Science Fiction, when it is originally written, seems like just that, FICTION. Not too many generations ago space flight for instance was fiction. We went from Flash Gordon to landing men on the moon in less than thirty years. There are many other science fiction themes that have become reality. 1984 for example.

The one that bothers me is Colossus The Forbin Project. If anyone is not familiar with the book or movie, it is about a computer that in the end, holds the world hostage. Another one would be The Matrix. How will we know when we are living in the real world or in a simulation. In my mind, AI could be the path to both of these so called "science fiction", stories.

 

[Djjerme]

Or, as Dr. Ian Malcolm put it: “Your scientists were so preoccupied with whether they could, they didn’t stop to think if they should.”

And therein lies reality, in my opinion. Science Fiction, when it is originally written, seems like just that, FICTION. Not too many generations ago space flight for instance was fiction. We went from Flash Gordon to landing men on the moon in less than thirty years. There are many other science fiction themes that have become reality. 1984 for example.

The one that bothers me is Colossus The Forbin Project. If anyone is not familiar with the book or movie, it is about a computer that in the end, holds the world hostage. Another one would be The Matrix. How will we know when we are living in the real world or in a simulation. In my mind, AI could be the path to both of these so called "science fiction", stories.

Actually, my wife and I lean more toward Wall-E as our future. Look around, you already have everyone staring at a screen 99.9% of the day. Taking a trip to Disneyland or any large public venue, multitudes of people on their go go scooters. All we need is Buy N Low style conglomerate... (Amazon? Costco?)

 

[Martino]

It seems only yesterday (according to my parents and grandparents) that we were being told that everyone was watching too much TV, and before that listening to the wireless too much, or spending all their time at the movies/pictures/flicks. No doubt at one time there was criticism that people spent too much time with their nose stuck in books, or indeed parchments or ‘the scrolls’. (I say Claudius, have you got the scrolls? No, i always walk like this).
Plus ca change……

 

[Rhinochugger]

One thing I mention in the article — and something that shows up in examples like the first post — is that AI is very good at trying to make everything work. Sometimes too good. It can be overly encouraging, and it rarely stops you and says, “Maybe this is a bad idea.” If you truly believe you can fly, AI may help you design the wings instead of pulling you back from the cliff edge.

So the important part is to step back from the result, look at it with your own experience, and ask whether it actually makes sense.

Or, as Dr. Ian Malcolm put it: “Your scientists were so preoccupied with whether they could, they didn’t stop to think if they should.”

I like that bit

 

[korm kormsen]

this pic is an example, why i fear A.I.



these things react quickly, but don't notice, when they are wrong.

(lately i read, that military is planning to use A.I. to counter the new high speed rockets)

 

[GAP]

this pic is an example, why i fear A.I.

View attachment 355551

these things react quickly, but don't notice, when they are wrong.

(lately i read, that military is planning to use A.I. to counter the new high speed rockets)

Jake the Peg???

 

[korm kormsen]

Jake the Peg???

i mean producing outputs, that are not controlled by common sense...

 

[FatherMcD]

this pic is an example, why i fear A.I.

View attachment 355551

these things react quickly, but don't notice, when they are wrong.

(lately i read, that military is planning to use A.I. to counter the new high speed rockets)

As long as the rockets have only three fins, should be fine.

 

[GAP]

i mean producing outputs, that are not controlled by common sense...

An Obituary printed in the London Times​

Today we mourn the passing of a beloved old friend, Common Sense, who has been with us for many years. No one knows for sure how old he was, since his birth records were long ago lost in bureaucratic red tape. He will be remembered as having cultivated such valuable lessons as:

• Knowing when to come in out of the rain;
• Why the early bird gets the worm;
• Life isn't always fair; and
• Maybe it was my fault.

Common Sense lived by simple, sound financial policies (don't spend more than you can earn) and reliable strategies (adults, not children, are in charge).

His health began to deteriorate rapidly when well-intentioned but overbearing regulations were set in place.

Reports of a 6-year-old boy charged with sexual harassment for kissing a classmate; teens suspended from school for using mouthwash after lunch; and a teacher fired for reprimanding an unruly student, only worsened his condition.

Common Sense lost ground when parents attacked teachers for doing the job that they themselves had failed to do in disciplining their unruly children.

It declined even further when schools were required to get parental consent to administer sun lotion or an aspirin to a student; but could not inform parents when a student became pregnant and wanted to have an abortion.

Common Sense lost the will to live as the churches became businesses; and criminals received better treatment than their victims.

Common Sense took a beating when you couldn't defend yourself from a burglar in your own home and the burglar could sue you for assault.

Common Sense finally gave up the will to live, after a woman failed to realise that a steaming cup of coffee was hot. She spilled a little in her lap, and was promptly awarded a huge settlement.

Common Sense was preceded in death, by his parents, Truth and Trust, by his wife, Discretion, by his daughter, Responsibility, and by his son, Reason.

He is survived by his 4 stepbrothers;

• I Know My Rights
• I Want It Now
• Someone Else Is To Blame
• I'm A Victim

Not many attended his funeral because so few realized he was gone. If you still remember him, pass this on. If not, join the majority and do nothing.

 

[Madman]

Actually, my wife and I lean more toward Wall-E as our future. Look around, you already have everyone staring at a screen 99.9% of the day. Taking a trip to Disneyland or any large public venue, multitudes of people on their go go scooters. All we need is Buy N Low style conglomerate... (Amazon? Costco?)

And if you think about it, robots have been controlling our lives for a very long time. Think of Traffic Lights.....

 

[GAP]

And if you think about it, robots have been controlling our lives for a very long time. Think of Traffic Lights.....

SWMBO had a knee replacement just a couple of months ago and the surgeon used a robot to do that.
Simplest robot that I know of is a 3D printer.
Contrary to what Hollywood would have you believe computers do not think, they are only as good as the information we put into them.

 

[Madman]

SWMBO had a knee replacement just a couple of months ago and the surgeon used a robot to do that.
Simplest robot that I know of is a 3D printer.
Contrary to what Hollywood would have you believe computers do not think, they are only as good as the information we put into them.

That may be the present situation, but we're not far away at all from self thinking, self repairing and even self building computers.