Stop Externalizing the Cost of Your AI Use to Me: Collaboration Etiquette and Accountability in the Age of AI

A Blunt but Legitimate Complaint

A post titled Stop Externalizing the Cost of Your AI Use to Me recently made it to Hacker News. It didn't generate a massive discussion, but it touched on an increasingly sharp tension in modern technical collaboration: when someone uses AI to rapidly generate large volumes of content, the cost of reviewing, verifying, and correcting that content is quietly shifted onto the recipient.

The core argument is straightforward, yet it cuts right to the heart of the matter. Generating content has become unprecedentedly cheap — in seconds, a single prompt can produce a lengthy email, a detailed PR (Pull Request) description, or a seemingly authoritative technical analysis. But for the person reading it, the cost of verifying whether that content is accurate, truthful, or even useful hasn't dropped at all. In fact, it's often gone up, thanks to the sheer volume of words produced.

What Is "Cost Externalization" in AI Collaboration?

"Externalizing cost" is an economics concept referring to when one party offloads the costs of their activity onto a third party who had no say in the decision. The classic example is industrial pollution — a factory saves on environmental controls while nearby residents and society at large bear the consequences.

Three "Pollution" Scenarios in AI-Generated Content

In the context of AI-assisted work, this kind of "pollution" shows up in concrete ways:

• Ballooning code review burden: Someone submits a thousand-line Pull Request generated by AI, accompanied by a smooth but hollow description. The submitter may have spent five minutes on it; the reviewer spends hours checking logic line by line, verifying edge cases, and figuring out which parts are AI-hallucinated pseudocode.
• Bloated emails and documents: Something that could be said in three sentences gets "polished" by AI into three paragraphs of elaborate prose. The reader has to wade through the padding to extract the actual information.
• False sense of expertise in technical discussions: In issues or forums, someone pastes a lengthy AI-generated "analysis" that looks professional but hasn't been thought through or verified by the person posting it — leaving the community to sort out what's real and what isn't.

The problem isn't that AI was used. The problem is that the person using it isn't taking responsibility for what they produce. They kept the convenience of "generating" for themselves and handed the drudgery of "verifying" to everyone else.

Why AI Cost-Shifting Is Getting Worse

Generative AI has fundamentally changed the economics of content production. In the past, writing a thorough document or a well-reasoned block of code required genuine intellectual effort — and that effort served as a natural quality filter. People who were willing to invest the time generally had a basic handle on what they were saying.

That filter has been removed. When the marginal cost of generation approaches zero, anyone can produce massive amounts of content at minimal expense. But the attention and verification capacity on the receiving end remain finite and scarce. When cheap supply collides with expensive digestion, the scales tip.

Open Source Communities Are Hit First

This dynamic is especially visible in open source projects. Maintainers of several well-known projects have publicly complained about receiving floods of low-quality, AI-generated PRs and bug reports — submissions that look tidy on the surface but are untested, or simply don't work. Maintainers are already unpaid volunteers; now they're expected to also sift through "AI slop." curl maintainer Daniel Stenberg has been direct about this: AI-generated bogus security reports are wasting his team's valuable time.

Not Anti-AI — Just Pro-Accountability

It's worth being clear: this isn't an argument against using AI. The value of AI as a productivity tool is beyond question. The real ask is about collaboration etiquette and accountability: if you used AI to generate something, you're obligated to review it, verify it, and trim it yourself first — to make sure it's worth someone else's time.

A Simple Self-Check

One straightforward question can serve as a gut check: Have I personally read and vetted what I'm submitting? Am I prepared to stand behind every claim in it?

If the answer is no — if you can't be bothered to read through your own AI output — why should anyone else take it seriously? Dumping undigested AI-generated content on other people is fundamentally disrespectful: disrespectful of their time, and disrespectful of the collaborative relationship itself.

We Need a New AI Collaboration Contract

Every major leap in technology brings a lag in social norms, followed by a slow reconstruction of them. After email became ubiquitous, we gradually developed implicit etiquette around CC fields, reply-all, and appropriate length. When instant messaging took off, we developed unspoken norms around response times and message granularity.

Generative AI needs its own set of collaborative norms. A few core principles might include:

• Transparency about AI use: Be upfront when content was AI-assisted, so the recipient can calibrate their expectations.
• Human oversight as a requirement: AI produces the draft; a human is the final editor. Before you send something, it must pass through your own understanding and judgment.
• Brevity as a virtue: Don't let cheap generation become an excuse for bloated output. Respect your reader's attention budget.
• Internalize the cost: Whoever benefits from the convenience of generation should bear the responsibility of verification — not pass it on to others.

Closing Thoughts

Stop Externalizing the Cost of Your AI Use to Me resonates because it names a quiet pain point in AI-era collaboration: technology has made "producing" extraordinarily cheap, but it hasn't made "being accountable" any easier.

The most mature AI users aren't those who produce the most — they're the ones who take responsibility for what they produce. In an era where anyone can generate, the willingness to stand behind quality has become a rare and genuine virtue.

Don't externalize your costs onto me. That blunt line might be the best place for every AI collaborator to start their self-reflection.