The Elements of Style: Is Strunk and White Still Relevant in the AI Age?
William Strunk Jr. published The Elements of Style in 1918. His student E.B. White revised it in 1959. It has sold over 10 million copies. It is assigned in nearly every college writing class in America. It is 85 pages long and has more influence on English prose than any other book its size.
It is also, depending on whom you ask, either the greatest writing guide ever written or a "toxic mix of purism, atavism, and personal eccentricity." That second quote is from Geoffrey Pullum, a professor of linguistics at the University of Edinburgh. He is not alone in his criticism. But Strunk and White endures. The question for 2026 is whether it still matters - and whether AI changes the answer.
The short version: some rules are more relevant than ever. Some should have been retired decades ago. And one rule in particular turns out to be the single most useful principle for editing AI-generated text.
What Still Holds
"Omit Needless Words"
Rule 17 in the original. Three words that contain an entire philosophy of writing. Strunk wrote: "Vigorous writing is concise. A sentence should contain no unnecessary words, a paragraph no unnecessary sentences, for the same reason that a drawing should have no unnecessary lines and a machine no unnecessary parts."
This rule was important in 1918. It is critical in 2026.
A study of physicians using AI to draft patient messages found that AI-generated text was 5.8 times more verbose than human-written text for the same content. The humans wrote 254 characters. The AI wrote 1,470. Same information. Nearly six times the words.
This is not a healthcare-specific problem. It is an AI problem. Language models are trained on the internet, which over-represents formal, padded, qualification-heavy prose. They hedge. They add "It's important to note that" and "It's worth mentioning that" before every claim. They use three words where one would do. They turn "use" into "utilize," "help" into "facilitate," and "now" into "at this point in time."
Strunk's rule cuts through all of this. If a word does not earn its place, cut it. If a sentence says what the previous sentence already said, cut it. If a paragraph restates its own introduction in the conclusion, cut the conclusion. The principle is simple. Applying it to AI output is where the real work begins.
Use the Active Voice
Strunk's preference for active voice is one of the most repeated writing rules in English. "The active voice is usually more direct and vigorous than the passive," he wrote. "I shall always remember my first visit to Boston" is better than "My first visit to Boston will always be remembered by me."
AI loves the passive voice. "It should be noted." "It can be argued." "The results were found to be." The passive lets the model avoid committing to an actor. Nobody is noting. Nobody is arguing. Nobody found anything. Things just happen, as if by consensus of the universe.
As Orwell observed, the passive voice is a way to avoid responsibility. When Strunk and White tell you to prefer the active voice, they are telling you to name who did what. This forces precision. "Researchers found that AI text is 5.8 times more verbose" is clearer than "It was found that AI-generated text tends to be significantly more verbose." The active version names the actor, specifies the number, and drops the hedge.
Be Specific and Concrete
Prefer the specific to the general, the definite to the vague, the concrete to the abstract," Strunk wrote. Use "It rained every day for a week" instead of "A period of unfavorable weather set in.
AI gravitates toward abstraction. Ask it about a company's problems and you get "challenges in the current market landscape." Ask it about a person's qualities and you get "a passionate and driven individual." These sentences contain zero information. They could describe anyone or anything. They take up space without illuminating anything specific.
The fix is Strunk's: replace every abstract claim with a concrete one. Not "significant revenue growth" but "revenue grew 34% in Q3." Not "a talented engineer" but "she rewrote the search indexer in two weeks and cut query time by 80%." Specificity is not decoration. It is the difference between saying something and saying nothing.
Use Parallel Construction
"Express coordinate ideas in similar form," Strunk wrote. If you start a list with a verb, every item should start with a verb. If the first clause uses the present tense, so should the second.
AI breaks parallel construction constantly. It will start a list with a noun phrase, switch to a gerund, then shift to an imperative. It does this because it generates token by token, not structure by structure. It has no global awareness that a list should be parallel. Each item is predicted independently from the preceding tokens, and the most probable completion often differs in grammatical form.
This is a small rule with a large effect. Parallel construction makes lists scannable and arguments clear. Breaking it makes prose feel disjointed even when readers cannot identify why. When you edit AI output, check every list and every compound sentence for parallelism. It is one of the fastest ways to make generated text read like it was written by a person.
What's Outdated
"He" as the Universal Pronoun
Strunk and White recommended using "he" as a universal pronoun when the gender of the subject is unknown. "The reader will absently rest his eyes on the wall." White defended this in later editions, calling the singular "they" an error.
This was already controversial in the 1970s. Today it is simply wrong. The singular "they" has been standard in English for centuries. Shakespeare used it. Jane Austen used it. The Associated Press Stylebook, the Chicago Manual of Style, and the APA Publication Manual all accept it. Clinging to "he" as a universal pronoun is not a grammar position. It is a refusal to acknowledge how English actually works.
Ironically, AI models already use singular "they" by default. On this point, the machines have better instincts than Strunk and White did.
Prescriptive Usage Rules That English Has Moved Past
Strunk and White insisted that "hopefully" can only mean "with hope" - as in "she looked at him hopefully." Using it to mean "I hope that" (as in "hopefully it won't rain") was condemned as incorrect. They also objected to using "contact" as a verb. "Do not contact people," they wrote. "Get in touch with them, look them up, phone them, find them, or meet them."
Language evolves. "Hopefully" as a sentence adverb has been standard for decades. "Contact" as a verb is so normal that objecting to it sounds absurd. These rules were not observations about clarity or precision. They were personal preferences dressed up as grammar.
This is the "personal eccentricity" part of Pullum's critique. He pointed out that Strunk and White frequently broke their own rules. They used passive voice while condemning it. They padded sentences while telling you to omit needless words. White once wrote "the best designers of the age" after telling readers to avoid "the" before abstract superlatives. The book does not practice what it preaches
- at least not consistently.
Pullum's Full Critique
Geoffrey Pullum published his now-famous takedown in 2009, titled "50 Years of Stupid Grammar Advice." His argument was not that the book was useless. It was that the book was unreliable. It gave advice it did not follow. It stated rules as absolutes that were actually matters of style. It confused grammar (how English works) with taste (how Strunk and White preferred English to work).
He wrote: "The authors cannot distinguish between active and passive voice. They give wrong examples. They state claims about grammar that are simply false." His strongest point was that a book purporting to teach grammar should at minimum get the grammar right.
Pullum is correct on the specifics. Several of Strunk and White's examples of "passive voice" are not passive at all. Their rule against beginning a sentence with "however" has no basis in grammar. Their insistence on certain usages reflects mid-20th-century preferences, not timeless principles of English.
But Pullum's critique, sharp as it is, does not cancel out what the book gets right. The principles of conciseness, specificity, and active voice are sound. The problem is that Strunk and White mixed good principles with arbitrary rules, and the book provides no way to tell which is which.
The AI Angle: Strunk as an Editing Framework
Here is the argument for why The Elements of Style matters more in 2026 than it did in 2006. In 2006, the book was advice for human writers generating their own prose. In 2026, most writers are editing AI-generated prose. The job has changed. The book's best rules turn out to be perfectly suited to the new job.
Start with "omit needless words." This is the single most useful principle for editing AI output. Every AI model over-generates. Every model pads. Every model uses three sentences where one would do. Strunk's rule gives you a clear test for every word in every sentence: does this word earn its place? If not, delete it.
The physician study puts a number on it. If AI text is 5.8 times more verbose than human text, then roughly 83% of AI-generated words are needless by Strunk's standard. That is not a small editing task. It is the primary editing task.
Now add the active voice rule. AI defaults to passive constructions because the training data is full of academic and corporate writing that uses passive voice. "It was determined" instead of "we determined." "Improvements were made" instead of "we improved." Strunk's rule - prefer active voice - gives you a systematic filter. Go through the AI draft, find every passive construction, and rewrite it with a named subject.
Then the specificity rule. AI loves generalities because generalities are statistically safe. "There are many factors to consider" is a sentence that can never be wrong because it says nothing. Strunk's rule forces you to replace it with something that could be wrong - and is therefore useful. "Three factors matter: cost, speed, and reliability."
Together, these three rules - cut needless words, use active voice, be specific - form a checklist that transforms generic AI text into something worth reading. They are not the only editing principles that matter. William Zinsser's On Writing Well, Verlyn Klinkenborg's Several Short Sentences About Writing, and Orwell's six rules all cover ground that Strunk and White miss. But as a starting framework, three rules from a book published in 1918 are remarkably effective against text generated by technology invented in 2017.
How to Apply This in Practice
The problem with applying Strunk's rules to AI output is not understanding the rules. It is the mechanics of applying them. In a chat-based workflow - paste text into ChatGPT, ask it to cut needless words, paste the result back - you lose track of what changed. Did it cut words you wanted to keep? Did it rewrite sentences beyond what you asked? You cannot tell without reading both versions side by side, line by line.
This is where inline diffs change the equation. Athens shows every AI edit as a visible change in your document. Red for deletions. Green for additions. You can accept or reject each change individually.
Tell the AI to omit needless words and you see exactly which words it flagged. Maybe it cut "It's important to note that" from the beginning of three sentences. Good - accept those. Maybe it also cut a qualifier you wanted to keep. Reject that one. The process takes seconds instead of minutes, and you never lose track of what changed.
Tell the AI to switch passive constructions to active voice and you see each rewrite. Some will be improvements. Some will be awkward. You decide, change by change. This is Strunk's rule applied with surgical precision rather than applied blindly.
This matters because Strunk's rules are guidelines, not laws. That is the lesson Pullum's critique teaches. Active voice is usually better. Not always. Conciseness is usually better. Not always. Sometimes the passive voice is the right choice. Sometimes a longer sentence serves the rhythm. The writer needs to see each change and judge each case. A tool that applies rules wholesale, without showing what it changed, violates the spirit of what Strunk was teaching: that every word is a choice, and every choice matters.
The Verdict on Strunk and White in 2026
The Elements of Style is not a perfect book. It mixes timeless principles with dated prejudices. It states opinions as facts. It breaks its own rules. Pullum is right that it should not be treated as gospel.
But its core principles - omit needless words, prefer active voice, be specific, use parallel construction - are sound. They were sound in 1918. They are sound in 2026. And they are more actionable now than ever because the primary writing task has shifted from generation to editing.
When you generate text with AI and then edit it, you are doing exactly what Strunk trained his students to do: looking at every sentence with fresh eyes and asking whether it earns its place. The difference is that Strunk's students were editing their own drafts. You are editing a machine's draft. The skills are the same. The volume is higher.
The writers who will do the best work in the AI age are not the ones who generate the most text. They are the ones who cut the most. Strunk told us this a hundred years ago. We just did not know yet how much cutting there would be.