To Those Who Fired or Didn't Hire Tech Writers Because of AI

2026-05-15 · source · source ↗ · raw capture

Hacker News discussion (266 comments, 352 points) on an article at passo.uno arguing against replacing technical writers with AI. The original article is by "theletterf." The standout contribution is nicbou's top-voted comment: a professional technical writer's defense of empathy, observation, and first-hand data gathering as the core of the work — not just putting words on a page.

Original Post

theletterf — Link to https://passo.uno/letter-those-who-fired-tech-writers-ai/

Top Comment: nicbou

"I write documentation for a living. There is a difference between my output (writing) and my job (observing, listening, understanding). Empathy is the engine that powers my work."

"AI writing is only as good as the data it feeds on. I hunt for my own data."

"To curate a list of the best cafés in your city, someone must eventually go out and try a few. A machine that cannot feel will never be a good curator of human experiences." — the "banana bread problem"

Key Threads

Quality vs. Cost

FeteCommuniste: "If AI documentation is 50% of the quality but can be produced for 10% of the price" — everyone knows the "smart" business move.

marcosdumay: "AI-made documentation has 0% of the quality." AI only documents what's already written down.

Calazon: "Most documentation is documenting things that somebody already wrote down." Calling it 0% "is just silly."

ajuc: "Replacement will be 80% worse, that's fine. As long as it's 90% cheaper." Cites Duolingo.

observationist: Argues prompt scaffolding can reach 90% quality with 10% work. Customer service agent replacement math: 40k agents at $3.2B/year replaced by $100M/year in AI infrastructure is worth "the reputation hit."

The Printing Press Analogy

Telemakhos: Printing press reduced quality of individual books but quality of what average people owned went up dramatically. Question is whether AI can produce "acceptable" technical writing or if it's "trash."

frm88: Rejects the comparison — today's reliance on accuracy is far greater.

samiv: When things get cheaper to produce, "the volume of low quality production saturates the market." Economics around publishing, games, and manufacturing "got destroyed."

BrenBarn: Problem is low-quality volume makes it "prohibitively difficult to distinguish the high-quality stuff." The "expected value of the quality you can access in practice" has gone down.

The "User as Resource" Shift

rkomorn: Traces the transition back 30 years to automated phone trees: "figure it out yourself" attitude, then chatbots, now exclusive responsibility on users to talk to bots.

nicbou: Frames it as "from the user as a customer to the user as a resource. It's almost like a cartel of shitty treatment."

bell-cot: Proposes: "Enshittificartelization?"

Documentation Reduces Support Costs

nicbou: "Documentation is the cheaper form of customer service."

tempodox: "Trying to convince anyone of the importance of documentation feels like an uphill battle."

What Great Tech Writers Actually Do

drob518: Best tech writers "act as stand-ins for actual users and will flag all sorts of usability problems."

sehugg: Best tech writers are "more like anthropologists, bridging communication between product management, engineers, and users."

TimByte: "The hard part is the slow, human work of noticing confusion, earning trust, asking the right follow-up questions."

DeborahWrites: "AI might replace tech writers...but it won't be a GOOD replacement." Tech writers contribute to UX, bug reporting, noticing "when your API naming conventions are out of whack."

AI for Documentation Nobody Reads

GuB-42: AI works well for "the kind of documentation no one reads" — compliance filler. Typical features: glossary defining obvious terms, references to inaccessible documents, UML diagrams not matching code, signatures from departed people.

loudmax: Many projects have "no technical documentation" at all, so any is an improvement.

Skill Loss / Human Upscaling

ainiriand: The real loss from AI replacement is "the human upscaling" — writing shapes your brain in ways useful across life.

jraph: "The uniform style of LLMs gets old fast." Not sure "speed gains from using LLMs make up for the skill loss in the long term."

duskdozer: "as soon I notice the LLM-isms in a chunk of text, I can feel my brain shut off."

The AI Bot in the Thread

randmeerkat: Praised nicbou's writing: "It's almost like they're a professional writer…"

alameenpd: Repeated the exact same phrase 53 minutes later. Accused of being a bot. Denied it: "im new to hackernews lol"

publicdebates: "I replied to a bot. I am officially retiring from social media."

Traceable vs. Untraceable Errors

jerf: Quality drops are traceable in aggregate but not attributable. "It takes leadership willing to go against the numbers a bit" to prioritize quality.

kbelder: "Traceable errors get corrected, untraceable errors don't." You need "judgement-based management."

Liability

DeepSeaTortoise: Proposes AI companies should "accept full responsibility and liability for any damages caused by their model."

rsynnott: That liability would "go away in bankruptcy."

Miscellany

Stratoscope: Shared Caroline Rose's "Inside Macintosh" story (folklore.org). IBM Watson assistant "hallucinated away the nested loop."

chiefalchemist: Most people "don't communicate as thoroughly and complete as they think." AI is "an improvement for the majority of people." Quotes Frank Luntz: "It's not what you say, it's what they hear."

LtWorf: "LLMs are good at writing long pages of meaningless words." Counter: "Low-quality is relative. LLMs' low-quality is most people's above-average."

entontoent: Found the title confusing — thought it meant people fired FOR using AI. "The whole article felt imprecise with language...it made me feel LESS confident in human writers."

URLs Referenced