I’ve been noticing something annoying lately. Open any technical document, business proposal, or blog post generated by artificial intelligence, and you’ll see bold text everywhere. It’s like someone took a highlighter to every third sentence and decided that was good communication. But here’s the thing: when everything is emphasized, nothing is emphasized.
The tools we have for typographical emphasis are pretty straightforward. Bold, italic, capitals, and underlines. Each has its place, but bold is getting absolutely hammered right now. I suspect AI language models picked up this habit from their training data and amplified it, creating a feedback loop where more and more text gets the bold treatment because that’s what the models think good writing looks like.
Capitals have mostly fallen out of favor, and for good reason. When you READ TEXT IN ALL CAPS it feels like someone is SHOUTING AT YOU. It’s been widely recognized as the typographical equivalent of being that person at the party who talks too loud. Underlines disappeared once hyperlinks took over the web, they’re now permanently associated with clickable text. Both of these forms of emphasis also carry a bit of a stigma from the typewriter era, when they were the only options available beyond just writing normally.
The Subtle Power of Italics
Italics are my go-to for emphasis, though I use them sparingly. They don’t jump out when you’re skimming a page. Instead, they make their presence felt when you’re actually reading the text, following along with the flow of ideas. This is exactly what I want most of the time. When I italicize a word, it’s because I want you to hear that slight vocal stress if you were listening to me speak this paragraph out loud.
I write the way I talk, and italics help convey that spoken rhythm without being aggressive about it. They’re the typographical equivalent of leaning in slightly during a conversation to stress a point, not raising your voice.
When Bold Actually Works
Bold has one superpower that nothing else can match. It catches the eye during a skim. When I’m glancing over a long document looking for something specific, bold text acts like little signposts. This is why headings are almost always bold. They need to help readers navigate, to find the section they’re looking for without reading every word.
But this superpower only works if it’s used sparingly. Scatter bold text throughout your paragraphs and the reader’s eye doesn’t know where to land. Everything becomes noise. The navigation system breaks down.
There’s one specific use of bold within prose that I actually like. When I introduce an unfamiliar term and then explain it, I’ll bold it at the point of explanation. I picked up this technique from an AI textbook by Giarratano and Riley years ago. The idea is simple: when that weird term shows up again twenty pages later and you’ve forgotten what it means, you can glance back through the pages and the bold catches your eye immediately, reminding you where the explanation lives.
The key detail here is bolding at the point of explanation, not necessarily at first use. Sometimes a term appears first in a list like “We handle three processes: frobning, gibbling, and eorchisting.” You don’t bold them there. You bold them later when you actually explain what the hell eorchisting means.
The Callout Alternative
I see a lot of writers trying to make important sentences stand out by bolding the entire sentence within a paragraph. The intent makes sense, but there’s almost always a better tool for the job: callouts. You know those boxes or highlighted sections that sit beside or within the main text? They do everything bold sentences try to do, but better.
Callouts naturally draw the eye even more effectively than bold. They also give you the freedom to reword things specifically for skimmers. You’re not constrained by making the text work both as part of the prose flow and as a standalone highlighted statement. You can optimize for both contexts separately.
The Bullet Point Problem
Another marginal case: bolding the first few words of each bullet point in a list, treating them like mini-headings. I get why people do this. It’s trying to make the list more scannable. But honestly, the bullets themselves already do that job. Your eye naturally catches on those little dots or dashes.
And while we’re at it, bullet lists themselves are overused. Most of the time, a well-written prose paragraph flows better and is more pleasant to read than a stack of bullets. Writing isn’t just about conveying information efficiently, it’s about making the experience enjoyable for your reader. Bullets feel mechanical. Prose feels human.
The Feedback Loop We’re Stuck In
The real problem with AI-generated text isn’t just that it overuses bold. It’s that this writing style is now being fed back into training data for future models. People see AI-written text with excessive bold, they mimic it in their own writing, and that becomes part of the corpus that trains the next generation of models. We’re in a typographical race to the bottom.
I was tempted while writing this to create a deliberately awful paragraph with bold scattered everywhere, just to show you how ridiculous it looks when every third word demands attention. But I realized that would undermine the actual explanation I’m trying to give you. So imagine it instead: a paragraph where important words and key phrases and critical concepts all fight for your attention until your eyes glaze over and you stop reading entirely.
That’s what we’re doing to our readers when we treat bold like a magic wand that makes writing better by sheer force of emphasis. Typography works through contrast and restraint, and right now we’re losing both.