Bizarre Northern Powerhouse rail map issued

[TUC]

I'm going to have to take issue with that, as a former member of the WYCA comms team (which covers more than just the Mayor‘s office). If it came from DfT it would have been included without question (because they always know what they’re doing, write?). If it was done in-house it wouldn’t have been thrown in without someone else looking at it.

We don’t know where it came from and who signed off on it, but blaming a “junior” is wrong.

Fair enough, but it does throw the sign-off process into question.

 

[MarkyT]

View attachment 196020
image shows a map shared on social media by the Mayor of West Yorkshire (can't link as don't have an account) showing potential Northern Powerhouse Rail upgrades. From west to east the stations are:
Liverpool, Warrington (connections to Chester), Manchester Airpont, Manchester Airport, Warrington, Warrington, Huddersfield, Huddersfield, Manchester Piccadilly (located just south of Sheffield), Sheffield (connections to Hull which is placed on the map somewhere north of Selby), Rotherham, Leeds, York, Hull (again) (just north of York), York (again) (just north of the New Hull), Newcastle (somewhere near Filey)

The original image/post was removed quite quickly, I understand. ISTR a favourite Noel Dolphin comment: 'needs more Warringtons'.

 

[317 forever]

She's the Mayor of West Yorkshire!!

Even so, she might have got her inexperience from her time at Westminster

 

[Sun Chariot]

The original image/post was removed quite quickly, I understand. ISTR a favourite Noel Dolphin comment: 'needs more Warringtons'.

Ah - I've just noticed all the marked destinations! Good spot @321362

 

[J-2739]

That is blatantly AI. Have you seen how the Manchester Airport and Liverpool texts have been messed up?

 

[norbitonflyer]

Even the improved one seesm to have a quirk, as last time I looked Rotherham was on the existing route between Sheffield and Hull

 

[RichJF]

You can pay 1000s for top tier AI, but it will ALWAYS get geography wrong.
You use AI to streamline monotous processes & allow humans to do more engaging activities.

 

[YorksLad12]

Fair enough, but it does throw the sign-off process into question.

Absolutely. You and I (and others) would have pointed out the map was unbelievably wrong, but might have been told to post it anyway. If it really was AI generated, then the person who drafted the prompt needs some education too.

I know we're not perfect, but I do get a bit defensive about my profession sometimes

 

[Djgr]

Reminds me of my RUCC days, when I pointed out to First North Western all the "typos" on the map on their website and was basically told (aggressively) that they weren't interested.

 

[Sonik]

The lazy, poop-from-a-butt AI generated one was removed and one was then uploaded which had been produced by a person, probably in a panic.

I'm going with the theory that it was possibly done deliberately, to generate viral publicity for the NPR announcement.

After all it's so badly wrong that anyone with even a cursory knowledge of the geography should have spotted it, and presumably Brabin's team are locals.

The dodgy map has been widely shared online, including by many who are not rail nerds, so (intended or not) it has publicized the scheme to a much larger audience.

 

[Sun Chariot]

I'm going with the theory that it was possibly done deliberately, to generate viral publicity for the NPR announcement

If it was thus, then - in this DEI age - it's very poor judgement and counter-effective for some.

My 24 year-old son is dyspraxic, autistic and has day-to-day significantly impaired cognitive development. It was a huge step for him in his adulthood, to learn and then to remember bus numbers, routes and stops for the buses he relies upon. His local bus route map is invaluable to him. He doesn't at all understand the relationship between all its locations but he recoghises road names of the stop(s) he must disembark at.

He takes things at face value; he does not understand errors such as this map's.

An intentionally-erroneous map would result in his confusion and, based on times before, his getting off any stop in panic, in tears and unaware of his location.
Cue, dad and mum driving the entire route, to locate him.

 

[Bantamzen]

I'm going with the theory that it was possibly done deliberately, to generate viral publicity for the NPR announcement.

After all it's so badly wrong that anyone with even a cursory knowledge of the geography should have spotted it, and presumably Brabin's team are locals.

The dodgy map has been widely shared online, including by many who are not rail nerds, so (intended or not) it has publicized the scheme to a much larger audience.

If that's the thinking in the inner circles of the industry, then it really is screwed quite honestly.

However I suspect rather than be deliberate, it's just someone being lazy and using AI to bring up the map for them to send out on social media.

Just in a this case, AI made up an approximation of the map rather than the actual one.

I've started to use AI often in my work, and it is very, very easy to get erroneous results if you aren't specific enough, and why it's vital to QA anything that you use AI for before committing it to the real world.

 

[Tetchytyke]

As a proud Bradfordian, I can say with certainty that I don't want to live in a world where there are two Huddersfields. One Huddersfield is one too many.

== Doublepost prevention - post automatically merged: 19 Jan 2026 ==

I've started to use AI often in my work, and it is very, very easy to get erroneous results if you aren't specific enough

The issue with AI is that it doesn't have the I, but people treat it as though it does. I don't blame people for this, the techbros who are pushing AI advertise it as doing a lot more than it really does.

I use it in my work as initial investigation, as a way of distilling a list of search engine results into something a bit more usable. But it hallucinates all the time. If I am searching for a particular company it will often give me information about a company with a similar name.

When I correct it, it will sometimes then give me the information about the company I am looking for. But sometimes it will instead give me the information about the company with a similar name but just tell me that it is actually the company I am searching for.

The problems come when people don't double-check the AI output.

 

[Scanderina]

NPR is all about bringing Manchester and Yorkshire closer together after all

 

[Bantamzen]

As a proud Bradfordian, I can say with certainty that I don't want to live in a world where there are two Huddersfields. One Huddersfield is one too many.

I suppose they could be combined into a wider area of CleckHuddersFax, but yeah two is far too many...

The issue with AI is that it doesn't have the I, but people treat it as though it does. I don't blame people for this, the techbros who are pushing AI advertise it as doing a lot more than it really does.

I use it in my work as initial investigation, as a way of distilling a list of search engine results into something a bit more usable. But it hallucinates all the time. If I am searching for a particular company it will often give me information about a company with a similar name. When I correct it, it will sometimes then give me the information about the company I am looking for. But sometimes it will instead give me the information about the company with a similar name but just tell me that it is actually the company I am searching for.

The problems come when people don't double-check the AI output.

AI is very good for automating basic processes, especially when it's sandboxed away from the noise of the rest of the InterWeb. But the standard AI tools used by most have to be fact checked, it's so easy to ask something that makes sense to another human but will be vague to AI. I've seen some very odd outputs from it when I've not been very specific in an ask.

 

[Pigeon]

The correct name for the technology concerned is Artificial Stupidity and it would help a LOT if people would remember that.

 

[Agent_Squash]

The correct name for the technology concerned is Artificial Stupidity and it would help a LOT if people would remember that.

I don't think that's a reasonable assessment. The issue isn't necessarily the technology itself - which, if used properly, is a game changer - it's the person sat between the keyboard and the chair a lot of the time.

A lot of times these days AI is being used as an excuse for laziness.

 

[dosxuk]

I suspect it's far more likely someone got AI to draw up a placeholder image, which was then passed to a human designer to draw properly - but the placeholder got accidentally attached rather than the final version. This sort of thing happens all the time, the only real difference is how much of a hash AI can make of things while still looking like it was intended to look.

This is not helped by an increasing number of people who can't sign off design work which has sample text or "PLACEHOLDER" written on images because "it doesn't look like the final image", so you end up with bad mock ups and nonsense, but plausible text, floating around just waiting to get accidentally distributed.

 

[Pigeon]

I don't think that's a reasonable assessment. The issue isn't necessarily the technology itself - which, if used properly, is a game changer - it's the person sat between the keyboard and the chair a lot of the time.

A lot of times these days AI is being used as an excuse for laziness.

That's part of the point really. That problem could be significantly reduced if the tool in question, rather than being deliberately given an inappropriate and misleading name chosen for its use in science fiction to denote something far more sophisticated, was instead named to reflect its true nature as a highly proficient computerised bull artist.

As a recourse of laziness it fails because of the need to check that the output is actually correct rather than just looking plausible, which is likely to be a much more difficult task than whatever the original assignment was in the first place.

 

[TUC]

The puzzling aspect is that it's hardly a complex map to draw manually in the first place.

 

[Ian Hardy]

And they call it AI, (Amazing Incompetence), which, we are told, is the future of everything.

 

[weeklycommuter]

The really worrying thing is that trillions of US dollars are being invested in AI. What happens when the penny drops that it is all useless?

 

[JohnMcL7]

The really worrying thing is that trillions of US dollars are being invested in AI. What happens when the penny drops that it is all useless?

There are significant concerns about what happens when the AI bubble pops with it increasingly looking like it's going to do a lot of damage because many big companies are heavily invested in AI (in some cases with far more money than they actually have) with the expectation that the LLM performance will scale with much more hardware. There are significant flash memory shortages which are going to put up the prices of almost everything electronic as companies divert resources to manufacture parts for AI data centres.

The image in the first post in the thread does highlight one of many problems with AI because while there's a few people saying it should have been checked which is true, I just find it's easier to do the work myself and then I know it's correct. Even if the AI output is correct, it still needs carefully checked because there's such a high probability there are errors.

 

[BigB]

The image in the first post in the thread does highlight one of many problems with AI because while there's a few people saying it should have been checked which is true, I just find it's easier to do the work myself and then I know it's correct. Even if the AI output is correct, it still needs carefully checked because there's such a high probability there are errors.

and here is where many issues are compounded - a lot of people just check if things are correct using - you guessed it - an AI generated summary that compounds the errors… there is no substitute for taking the time to check it yourself but that’s not the world that the tech giants want us to live in.

No conspiracy theories, tin foil hats or blue Monday hangovers were used to make this view!

 

[90019]

and here is where many issues are compounded - a lot of people just check if things are correct using - you guessed it - an AI generated summary that compounds the errors…

Or they'll check the first few times and find that the output is good enough, so they stop checking because it obviously works fine.

 

[Smwrff]

In defence of AI - I use it a lot for fact gathering and preparing drafts - but never rely on it to produce final publishable douments. All output needs to be sense checked by a competent person.

This is represented by Murphys Law of AI:

AI & L = NS
Where L = Laziness
NS = Natural Supidity

Plenty of examples around on Linkedin.

 

[Retorus]

In defence of AI - I use it a lot for fact gathering and preparing drafts - but never rely on it to produce final publishable douments. All output needs to be sense checked by a competent person.

This is represented by Murphys Law of AI:

AI & L = NS
Where L = Laziness
NS = Natural Supidity

Plenty of examples around on Linkedin.

I’ll pass, thanks.

 

[Dave W]

The image in the first post in the thread does highlight one of many problems with AI because while there's a few people saying it should have been checked which is true, I just find it's easier to do the work myself and then I know it's correct. Even if the AI output is correct, it still needs carefully checked because there's such a high probability there are errors.

This may be true in your field but AI has made significant inroads in terms of efficiency especially at administrative level.

I wouldn't be using it to create images to publish, though.

 

[Grumpy]

This shambles and the shambles at the White Rose project doesn't give an image of a competent professional management in the WYCA.
Has anyone been told to clear their desk?
Thought not

 

[frodshamfella]

Liverpool Gateway Station, is not showing.