What is a dead verb?
A dead verb is a verb that isn't actually doing any work. Instead of naming the action directly, it leans on a helper word — was, began to, seemed to, could — and pushes the real verb off into the background. The sentence technically describes something happening, but the happening feels secondhand, like it's being narrated through a pane of glass.
They're called "dead" because the main verb has been emptied of force. It's alive grammatically but dead on the page.
Dead verbs aren't grammatically wrong. You won't find a red squiggle under them. That's exactly why they're dangerous — they pass every spellcheck and still make prose read like it came out of a machine.
The four families of dead verbs
1. The "was [verb]ing" construction
The classic. A past-tense be verb glued to an -ing participle. It's the construction AI reaches for more than any other.
The fix is almost always mechanical: drop was, convert the participle to simple past. Was sitting → sat. Was looking → looked. The sentence loses a word and gains momentum.
2. The delay verbs: began to / started to / proceeded to
These pretend to add nuance — "she began to cry" suggests a different moment than "she cried" — but in practice they almost never carry that nuance. They're filler. The character either does the thing or doesn't.
If you genuinely need the onset of an action, a physical beat does it better than "began to." "He opened his mouth, then stopped" shows the start and the abort; "he started to speak, then stopped" just labels them.
3. The hedge verbs: seemed to / appeared to
These are subtler, and that's the problem. They let the writer avoid committing to what's actually happening — which is precisely what makes them read as evasive, the verbal equivalent of shrugging.
The fix isn't to swap in a stronger word for "angry" — it's to drop the emotion label entirely and show the body. Hedges exist to hide behind; cut them and you're forced to commit.
4. The filter verbs: could feel / could hear / could see
These insert a layer of narration between the reader and the sensation. Instead of the reader feeling the cold, the reader is told that the character could feel the cold. Two steps removed.
Why AI writes this way
Large language models are trained on enormous amounts of text, and dead verbs are statistically common in prose — especially in summary, recap, and casual narration. When an AI generates fiction, it reaches for the most probable next word, and the most probable verb construction is usually a safe, hedged, participle-based one.
Human writers, by contrast, are dragged toward dead verbs by fatigue — they're easy, they fill a line, they feel done. A trained editor's job has always been to hunt them down. The difference now is that AI produces them by default, in volume, where a human only lapses into them when tired.
That's why dead verbs are the #1 fingerprint: not because they're uniquely AI, but because they appear at a density and consistency that no alert human writer produces.
How to fix them — the short rules
- "Was [verb]ing" → convert to simple past.
was sitting → sat - "Began/started/proceeded to [verb]" → cut the delay.
began to walk → walked - "Seemed/appeared to [verb]" → commit and show it physically.
seemed angry → his jaw tightened - "Could feel/hear/see [noun]" → drop the filter, put the reader in the sensation.
could feel the cold → the cold bit - When in doubt → ask whether removing the helper word changes the meaning. If it doesn't, the word was dead.
When to keep a dead verb
Not every dead verb should die. "She was sitting when the phone rang" carries information "she sat" doesn't — she was already seated when the interruption hit. The progressive aspect earns its keep when the ongoing-ness is the point: a state that was already in progress when something else broke in.
The goal isn't zero dead verbs. It's zero gratuitous ones — the ones that exist because the writer (or the model) reached for the path of least resistance rather than the verb the moment actually wanted.
A dead verb is a signal, not a sin. When you find one, don't just swap it — ask what the sentence is actually trying to do, then write the verb that does it.
Find every dead verb in your manuscript
Inkstride's Kill List scanner catches all four dead-verb families — plus 20+ other AI fingerprints — in seconds, with one-click fixes. Free to try.
Open the Editor — Free Try the Free CheckerFurther reading
Dead verbs are one category in a larger set of AI writing fingerprints. If you're cleaning up prose, the same logic applies to telling emotions instead of showing them, AI crutch phrases like "the weight of" and "a dance of", and overengineered dialogue tags.