Mystery coach

[jhb171achill]

Saw this on another FB page. What on earth is this thing on the right hand side? Almost looks AI-generated. Warrenpoint in UTA days, for context.

 

[David Holman]

Headlight, lots of hoses. Driving end of a railcar/dmu. Or inspection saloon maybe?

[Mol_PMB]

Knuckle coupler, no side buffers. Proportions all wrong. AI nonsense.

[DJ Dangerous]

Perhaps a screenshot of the original post including the poster might help.

[jhb171achill]

12 hours ago, Mol_PMB said:

Knuckle coupler, no side buffers. Proportions all wrong. AI nonsense.

Indeed - but it’s an old photo!

There certainly was never anything in existence that looked even remotely like this.

[Mol_PMB]

Your version has definitely been fiddled with by AI. Here is a view from a similar angle at a similar date, but not the same photo:

One of the things that AI often gets wrong is the track. Look at your photo and there are some odd kinks and gauge variations. Also there is a stray piece of rail (looking almost like a switch blade) in the track second from left. In reality there was a trap point blade near here (see Ernie's photo above) but it faced the opposite direction and of course it had some rodding to work it.

 

AI isn't afraid to make things up and claim it to be true. And then to lie to cover its tracks. It will become an increasing hazard to historians, as the internet fills up with lies and nonsense.

[DJ Dangerous]

1 hour ago, Mol_PMB said:

Your version has definitely been fiddled with by AI. Here is a view from a similar angle at a similar date, but not the same photo:

One of the things that AI often gets wrong is the track. Look at your photo and there are some odd kinks and gauge variations. Also there is a stray piece of rail (looking almost like a switch blade) in the track second from left. In reality there was a trap point blade near here (see Ernie's photo above) but it faced the opposite direction and of course it had some rodding to work it.

 

AI isn't afraid to make things up and claim it to be true. And then to lie to cover its tracks. It will become an increasing hazard to historians, as the internet fills up with lies and nonsense.

That’s why private and paywalled collections are going to be so valuable in the future.

We haven’t been given the context for this photo. Quite possible that the original poster was openly displaying their AI generated work and asking for opinions.

A different hobby, but this video analysis shows just how such initial honesty and openness can be twisted, misused or taken out of context:

 

 

[Auto-Train Original]

So many bogus historic photos on social media  are to train the AI. It uses the responses to try to determine by the comments what's is probably the most correct response by the number of like responses.

[jhb171achill]

2 minutes ago, Auto-Train Original said:

So many bogus historic photos on social media  are to train the AI. It uses the responses to try to determine by the comments what's is probably the most correct response by the number of like responses.

And that, in itself, is a scary concept.

 

[Broithe]

AI.

Artificial Ignorance.

Not quite as good as real ignorance yet, but it's getting there.

[skinner75]

AI can lie too - don't think it can't

[Auto-Train Original]

2 hours ago, skinner75 said:

AI can lie too - don't think it can't

It is pretty worrying when you read some of the reports about it telling lies and then being caught and it being deceptive or trying to gaslight its way out of being found out. It's not good and is unregulated. Military Photo Archives are being decimated by it with bogus AI images of conflicts/events. Shocking stuff.

[Broithe]

6 minutes ago, Auto-Train Original said:

It is pretty worrying when you read some of the reports about it telling lies and then being caught and it being deceptive or trying to gaslight its way out of being found out. It's not good and is unregulated. Military Photo Archives are being decimated by it with bogus AI images of conflicts/events. Shocking stuff.

It will become difficult, impossible, or just not worth the effort, to refute the nonsense that people 'know that they know'.

Only a couple of weeks ago, I saw posts where someone was stating widely repeated "facts" about an aircraft - they were virtually just propaganda nonsense, but he wouldn't accept the truth, even when it was being supplied by a chap who had actually flown the individual aircraft in the photo being discussed - and for many years.

It's not all down to AI, but it will greatly accelerate the 'fake fact' phenomenon.

Even before AI, people could be a bit dim - I remember being told that a clear fake picture couldn't be a fake - because it had been taken before Photoshop was invented.

I'm getting better at just leaving people to dwell in their own world...

[Mol_PMB]

https://www.forbes.com/sites/bernardmarr/2024/03/28/generative-ai-and-the-risk-of-inbreeding/

[Auto-Train Original]

Own and cherish tangible physical media. Personal hardcopy archives is our only choice at this stage. 

[Galteemore]

7 minutes ago, Auto-Train Original said:

Own and cherish tangible physical media. Personal hardcopy archives is our only choice at this stage. 

How Stalin would have loved it. So much easier than manual airbrushing. The issue is a significant worry, as surveys find people are putting more and more trust in social media (where AI thrives) as a news channel rather than established outlets. 

Edited Thursday at 11:34 by Galteemore

[Broithe]

1 minute ago, Galteemore said:

How Stalin would have loved it. So much easier than manual airbrushing 

He just needed to learn how to (ice)pick his friends more carefully...