The Illusion of Quality. Stop trusting pretty documents.

There’s a new trend emerging in teams where at least one member uses AI heavily. Documents are getting longer and more visually appealing. They come with dense technical language, many sections, and a confident tone. On the surface, they look extremely professional.

But surface can be misleading.

Broken signal

A well-written document has always been a signal of competence, perfectionism, and care. We learned to trust polish because polish used to cost effort. Someone who produced a structured, well-worded document had almost certainly spent time thinking through the problem.

AI broke that signal. It can wrap half-formed ideas in sophisticated vocabulary, organize scattered thoughts into convincing sections, and sprinkle words like comprehensive, robust, scalable, and strategic across every paragraph. The result is a document that looks definitive, even when there’s no underlying thinking. A pile of trash in a beautiful package.

The danger is not just that the content could be wrong. The danger is that it looks right enough to stop people from questioning it.

It happens to all of us. When we see something that looks sophisticated, we have an instinct to trust it. We nod along. We assume the author has done the thinking and move forward. Not because we don’t care, but because the polished document used to be a reliable signal. Our instincts haven’t caught up with the new reality.

Team cost

As these AI-polished documents circulate, people hesitate to challenge what looks thorough. Meetings shift from challenge and debate sessions to approval ceremonies, causing the bar to lower, without anyone noticing.

The speed doesn’t help either. AI workers generate content at an intimidating pace, and that creates a subtle pressure to move forward as soon as something looks “good enough.” The person who led the AI to produce the document still owns the responsibility: choosing the right format, asking the right questions, stress-testing the output. But when a polished draft appears in minutes, it takes real discipline to slow down and do that work.

There used to be a friction to writing a document from a blank page, a friction that forced us to think. You couldn’t produce a strategy doc without wrestling with the strategy. You couldn’t write a technical proposal without confronting the trade-offs. It’s good that the friction of wording is gone. But the friction of thinking should never go away. Even now that AI can speed writing up by a lot, the bottleneck remains unchanged: thinking.

Tips for avoiding creating slop

I’ll be honest. I’ve also been tempted by the shiny presentation AI made for me out of a few random thoughts. So these habits are as much a reminder to myself as they are advice:

• Write long drafts. Spend time bringing your reasoning to words. Not necessarily the perfect wording (AI can handle structure, organization, and vocabulary), but the thinking has to be yours. Dictation tools can help if staring at a blank page feels paralyzing.
• Cut ruthlessly. If a section doesn’t add value, remove it. If two sections say the same thing, merge them. Length is not a proxy for quality.
• Create visual aids. Tables, diagrams, sequence flows, data models… AI is excellent at producing these with tools like Mermaid and Markdown. And we as humans are excellent at reading and understanding them.
• Challenge every decision. Read through and challenge each thing you don’t fully understand or agree with. This is your work, not a stranger’s. If you can’t defend a choice in the document, rewrite it until you can.

In the end, producing good work still requires thinking. AI hasn’t removed that need, it has only made it easier to hide the absence of it.