Blogger publicly shames Ocado over AI-written customer service reply
Terence Eden got a sycophantic AI reply to a bug report, tanked Ocado's satisfaction survey, and received a more human-sounding follow-up — sparking an HN debate on AI in support.
What to know
- A blogger identified an Ocado customer service reply as AI-generated from its sycophantic opening and em-dash, then protested via a satisfaction survey.
- Ocado's follow-up apology read more human, but HN commenters disputed whether it was genuinely human-written or just better-disguised AI.
- The episode is framed as an example of the 'AI Resist List' idea discussed by Professor Kate Devlin at EMFCamp: publicly calling out poor AI use.
- Discussion centers on whether AI-authored replies to human complaints are inherently disrespectful, or whether speed and quality matter more than authorship.
The dispute Whether the problem is AI authorship itself or simply the quality/effort of the response, since several commenters suspect the 'better' human-sounding reply was also AI-written. · positions read across 11 posts and comments
Using AI to reply to a genuine human complaint is disrespectful and erodes trust, even if AI routing/summarizing is fine.
-
“I was so insulted by the AI response that I'll never do that again.”
jjgreen · Hacker News ↗
AI itself isn't the issue — reply quality and speed matter more than whether an AI composed it.
-
“I think that good customer service can use AI. It's not AI composed replies that makes the customer service bad.”
tln · Hacker News ↗
The 'more human' second reply was probably also AI-generated, just better disguised with a style guide, so AI detection by tone is unreliable.
-
“Default llm gened text can be easily detected, but if operator provide sufficient style guide - it is almost undetectable. The second response was also llm, just with little bit of style guide.”
hyperionultra · Hacker News ↗
Terence Eden Blogger (shkspr.mobi)Ocado British online supermarketKate Devlin Professor, discussed the 'AI Resist List' at EMFCamp
How it unfolded 2 developments, newest first · click a bar or a number to jump articlespostscomments
-
2
Hacker News discussion debates AI in customer service
The blog post reached Hacker News front page, drawing dozens of comments disputing whether AI-authored replies are inherently disrespectful or whether the real problem is reply quality rather than AI involvement.
“You're entirely right—having a technical glitch pop up when you're trying to organise your delivery is a real pain, so I've made sure to pass your feedback on to our technical and webshop team.”
— Ocado, first customer service reply · source -
The current worst customer service departments are the ones that never set up a website. Presumably easy to use customer service would be a net loss because it wouldbe used more? But they've happily added AI to their phone tree! It burns ever more of your precious human lifespan while they have even fewer minimum wage slaves in their call…
2 more of the top 3 · 11 posts in this stretch
-
Assuming the person is serious and not writing this tongue in cheek...I find it humorous is that all it takes for OP to think the response was written by a human is that it says it was written by a human.Both responses could have been from AI. Or none. Or just one. Just look at the comments on this thread, nobody can agree. And in the end, it's…
-
We are just a couple years away from companies realizing the first line of customer service can be a profit center and have the largest impact on brand likeability for minimal cost.Yes, when people use customer support as a glorified FAQ lookup an AI response is appropriate, but a few humans that care can positively impact 100 customers a day. AI…
-
-
1
Eden publishes 'Just a little AI resistance, as a treat'
Citing Professor Kate Devlin's AI Resist List concept from EMFCamp, Eden wrote up the exchange as an example of publicly complaining about poor AI use, arguing that routing and summarizing support tickets with AI is reasonable but replying to a human with AI is not.
“But using an AI to reply to a human is just disrespectful. It says the organisation values their customer's time at less than their own.”
— Terence Eden -
1 outlet first by HN Frontpage, 7d ago · read ↗
-
-
background
Ocado sends a second, more personal-sounding apology — A few days after the negative survey, Eden received a follow-up message apologizing directly for the earlier 'robotic' reply and thanking him for the diagnostic detail, which he judged to read more like genuine human writing.
-
background
Eden gives Ocado 1-star marks on the satisfaction survey — In the follow-up customer satisfaction survey, Eden gave the lowest possible score on every metric and complained explicitly about using AI to answer a genuine bug report.
-
background
Ocado sends a reply Eden identifies as AI-generated — The response opened with a sycophantic agreement, used an em-dash uncommon in British style, and otherwise read nothing like typical customer service prose, leading Eden to conclude it was AI-written.
-
background
Eden reports a calendar-invite bug to Ocado via feedback form — Unable to find a technical contact or bug-report channel, Terence Eden used Ocado's customer service feedback form to submit a detailed technical analysis of a malformed calendar invite issue.
What people are saying 7 voices from 1 site · best of 11 · verbatim
- Was the second, more human-sounding Ocado reply actually written by a person, or just a better-tuned AI output?
- Where is the line between acceptable AI use (routing, summarizing, triage) and unacceptable AI use (direct replies to complaints)?
- Nov 3
-
I used em-dashes -- routinely -- before David E. Rumelhart even wrote "Learning representations by back-propagating errors" in 1986. AI hysteria is wrecking language as much or more than AI itself is.
-
Even if the second response used AI, it at least seams like someone took the time to edit it that response thoroughly to make it more digestible. I think that's what OP, and myself, would at least expect in that situation.
-
Default llm gened text can be easily detected, but if operator provide sufficient style guide - it is almost undetectable. The second response was also llm, just with little bit of style guide.
-
As a grumpy misanthrope I welcome the trend of overly sycophantic AIs increasing the value of borderline rude thank you messages as a sign of actual human responses, increasing the demand for grumpy misanthropes on staff, and hope that AI can't emulate us well soon.
-
I think Rodney Brooke’s three laws of robotics apply here, particularly how the appearance of the bot promises its capabilities. It might have better for an obviously automated response to come back. No one might be listening but would not have been promising human empathy while faking human attention.
-
For as much as I like AI for (gestures broadly around the hacker news space), I am glad that I'm not alone in getting real upset at the endlessly incessant user-directed AI garbage, be it customer service emails, or ChatGPT generated event posters.
-
I never contact customer service with a complaint, I just don't use them again. But I have, very occasionally, contacted customer service with a "could you pass on my thanks to ..." for a particularly good product or service. The last a couple of months ago to Marks & Spencer for whoever it was who discovered their Len de l'Ell wine, which is…