Harringay, Haringey - So Good they Spelt it Twice!
Manor House station, Green Lanes exit (west side)
I wrote to TFL a few weeks ago to ask about progress in resolving the issues that have resulted in this exit being shut for a few months now. The reply is pasted below. Lengthy, but no clues about what’s actually happening to resolve it! Is anyone else concerned about this and do you have any ideas about what we can do to escalate it? It’s not the end of the world for me, but if you’re not the most mobile it must be really frustrating. Perhaps our councillors might have a view?
In case anyone is curious about some of the comments in the letter, a lot of it is summarising the points I made). So much so you’d almost wonder if there was some AI involved …
"Thank you for contacting us about the ongoing closure of the Green Lanes exit at Manor House Underground station, and I’m sorry for the continued inconvenience this has caused.
"I understand your concern that this exit has now been closed for several months, having previously been shut intermittently during heavy rain, and that there were noticeable issues such as a sewage smell and the area not feeling safe.
"Where an exit is closed for an extended period, it is usually because the underlying issue is more complex than routine maintenance. In cases like this, problems related to drainage, water ingress, or third‑party utilities can require specialist investigation and coordinated remedial works before the area can be reopened safely.
"Your feedback about the impact on passengers changing to buses towards Harringay and beyond, and the difficulty of having to access the station on the opposite side of Green Lanes, has been noted. I also understand your wider concern about what feels like a pattern of neglect, and your reference to similar issues at Turnpike Lane.
"Thank you for taking the time to raise this in detail. We recognise how important clear communication is when facilities are unavailable for long periods, and we appreciate your patience while this is reviewed."
Tags for Forum Posts: manor house station, tfl
Replies to This Discussion
-
Permalink Reply by Hugh
-
Well done for raising the issue and trhanks for posting.
It does look suspiciously like an AI-generated answer, either through automation or an employee running your ,eter through an AI engine.
-
Permalink Reply by Clive Carter
-
Yes, I agree; I assume this fact- or content-free response did not go from Tfl over the name of person.
For a particular subject at local level, its a variation on the municipal "under review" and at national level "under constant review". At institutional middle-level, its a review with thanks for patience.
-
Permalink Reply by Gordon T
-
The closure of the west side exit (exit 6) is not just inconvenient, it's hazardous.
Getting to the bus stop for the 29 bus towards Harringay means you have to use exit 7 opposite which brings you to the wrong side of Green Lanes. You then have to cross four lanes of traffic to an island, then three more lanes, two of which carry traffic coming up behind you, looking out all the time for 'scoff-laws' on bikes not complying with traffic lights.
If a 141 or a 341 will do, you are much safer taking exit 4 (past the ticket machines and along the passage - narrowed for another set of leaks!, then left) and wait at the bus stop outside the Co-op convenience store.
-
-
Alison: well spotted!
TfL, like other organisations, imagine that purely synthetic answers will not be recognised. Such answers add no new information. When a vendor on eBay presses the button for an AI description of their item, the waffle/ blandishment words typically generate nothing new.
Automated Complaint-handling: on the basis that it takes one to know one, I pasted your response to Anthropic's Claude and asked how likely it was generated by AI. Answer:
———
This is a great question and one I can give you a pretty informed analysis on, drawing on both my knowledge of AI-generated writing patterns and the broader context of how organisations like TfL handle customer correspondence.
My assessment: Very likely AI-generated, or at minimum heavily AI-assisted.
Here's why:
Strong indicators pointing to AI authorship:
1. Structural "sandwich" format. The response follows a textbook AI pattern: acknowledge → empathise → explain generically → note the feedback → thank and close. This formula is characteristic of how large language models are trained to handle complaints.
2. Hedged, non-committal language throughout. Phrases like "it is usually because," "can require specialist investigation," and "while this is reviewed" make no actual promises and commit to nothing. AI systems are often tuned to avoid commitments; a human agent with knowledge of the specific situation would more likely give a concrete update or timeline.
3. Suspiciously comprehensive recall. The response accurately reflects several specific details the resident raised — the sewage smell, the bus routes toward Harringay, the comparison with Turnpike Lane, the intermittent previous closures. A human agent handling dozens of cases daily would rarely weave all these specifics in so fluently. AI, given the original message, does this naturally.
4. Generic explanation dressed as specific. The paragraph about "drainage, water ingress, or third-party utilities" sounds informed but actually tells the resident nothing they couldn't have guessed. It's a plausible-sounding filler explanation — classic AI behaviour when it lacks real case data.
5. Tone and vocabulary. Phrases like "coordinated remedial works," "specialist investigation," and "facilities are unavailable for long periods" have the polished, slightly formal register typical of LLM output. No human customer service agent would naturally write "your reference to similar issues at Turnpike Lane" — that phrasing mirrors how an AI reflects back input text.
6. No named agent, no case reference. Though this alone isn't conclusive, AI-drafted responses often lack the personal signing-off style of a human rep.
What might suggest human involvement:
The response does correctly identify and name the specific exit (Green Lanes exit at Manor House), the specific onward journey impact (buses toward Harringay), and the Turnpike Lane comparison. A pure template AI without the original complaint in context couldn't do this — so either a human drafted it with AI assistance, or an AI was given the full complaint text to work from (the more likely scenario).
Bottom line: The response reads as almost certainly generated by an LLM (like ChatGPT or a similar tool) given the resident's original message as a prompt — possibly within a customer service platform that TfL or a contractor uses. It is empathetic-sounding, grammatically flawless, and completely non-committal — all hallmarks of AI customer service output. A human agent who actually knew the status of the Manor House closure would very likely have said something more specific.
Whether TfL officially uses AI for correspondence drafting is not publicly confirmed, but the pattern here is very recognisable.
-
Permalink Reply by Alan Stanton
-
Reading the Transport for London reply, I suspected a dialect of English which long long preceded AI.
My suspicion is Bureaubabble.
I learned one of its variants, working as a solicitor's office clerk in the last century.I may be wrong of course. Which we may discover, if they feed the AI a diet of Scottish Working class novels; plus U.S. poetry, and modern song lyrics. Causing it's future replies to become far less opaque.
In the meantime. I'll draw attention to your post, to a friend of mine, on their return from the time-hallowed midsummer ritual of pacing the pavements. Performing ceremonial customs - some dating from childhood - like pressing the bells and not running away.
-
-
I like the way Anthropic's Claude got in a dig at or sly mention of, their competitor, ChatGPT!
As if to hint that if Anthropic had been asked to generate an auto reply, Claude might have done better. Which it wouldn't!
-
Permalink Reply by Zena Brabazon
-
Hi Alison
Thanks for posting. It reads like AI! The issues you are raising about the station are really important and as a first step I'll contact Joanne McCartney as she is our Haringey GLA member and also the statutory Deputy Mayor. The station is in Hackney so will also copy in the Hackney GLA rep. There is also a Deputy Mayor for Transport so can copy him in. His name is Seb Dance. I think getting a political intervention will get a better and proper response from TfL.
Best wishes
Zena
Zena Brabazon
Labour and Co-Operative candidate
Harringay ward
-
-
Good luck in getting TfL to fix the problem; hope Zena’s suggestions will help. Having said that, the response may well be AI-generated but I’ve seen many very similar letters from customer “service” departments over decades, all just using a standard template that, as here, admits nothing and commits to nothing. These days, I always write to or e-mail a CEO if there’s a problem; no guarantee of resolution, obviously, but mostly they have staff who may take it more seriously for reputational reasons
© 2026 Created by Hugh.
Powered by
