A data analysis of how many marketers now write with AI, how much of the web is machine-assisted, how much time it saves, and why readers still trust a human byline more than the tool that helped write it.
Eighty-seven percent of content marketers already use AI to create or help create content, per a survey of 879 practitioners by Ahrefs — only 13% still work without it. That adoption shows up in the output: when Ahrefs scanned 900,000 newly published web pages in April 2025, 74.2% contained AI-generated text. The numbers below separate the content-marketing reality from the hype: where AI genuinely saves time, how teams actually edit its drafts, and why the trust gap with readers is the real constraint.
Key takeaways
- 87% of content marketers use AI to create content, and just 13% do not, per Ahrefs
- 74.2% of new web pages contain AI-generated text, but only 2.5% are pure AI — 71.7% are a human-AI mix (Ahrefs)
- 55% of marketers name content creation as their top AI use case, up 12 points year over year, per HubSpot
- Only 7% publish AI output unedited — 56% significantly revise or rewrite it, per HubSpot
- 36% of AI users write a long-form post in under an hour, versus 38% of non-users who take 2-3 hours, per Semrush
- Human-written content still wins position one — 80.5% probability versus 10% for AI text across 42,000 blog pages (Semrush)
- 53% of consumers distrust AI-powered search results, per a Gartner survey of 377 US consumers
How many marketers actually use AI to write?
Adoption is effectively the default. Ahrefs found 87% of content marketers use AI to create or assist with content, and Semrush reports 76% specifically use it to draft copy. HubSpot adds that 55% now rank content creation as their single top AI use case — a 12-point jump in a year.
The through-line across surveys is that AI writing moved from experiment to workflow in roughly two years. The interesting question is no longer whether teams use it, but how much of the draft they keep. Our companion AI adoption statistics show the same pattern across the wider workplace.
How much time does AI writing actually save?
The gap is real but narrower than vendor decks imply. Per Semrush, 36% of marketers who use AI say they finish a long-form blog post in under an hour, while 38% of those who don’t use AI report spending 2-3 hours on the same task. AI compresses the first-draft stage; it does not eliminate the research, editing, and fact-checking around it.
That efficiency is why satisfaction runs high: Semrush reports 71% of businesses are very satisfied with their AI writing tools and 79% say content quality improved. HubSpot found 74% of marketers believe AI enhances their productivity — a claim the time data broadly supports.
How much editing does AI content get before publishing?
Human oversight is the norm, not the exception. Per HubSpot, only 7% of marketers publish AI-generated text without any edits. A majority — 56% — significantly revise or completely rewrite the draft, and another 38% make minor tweaks first. Fewer than one in fourteen pieces ships straight from the model.
That editing discipline explains why Ahrefs found only 2.5% of new pages are “pure AI.” The dominant mode is human-led and AI-assisted: Semrush reports 64% of SEOs use exactly that workflow, and 23% still create content with no AI at all.
Does AI-written content rank as well as human writing?
Not at the top. Semrush analyzed 42,000 blog pages pulled from the top results for 20,000 keywords and found human-written content holds an 80.5% probability of taking position one, versus 10% for AI-generated text. Human writing kept an advantage across every top-10 position in the study.
The takeaway is not that AI content fails, but that unedited AI content struggles to reach the highest-value slots. The pages that win tend to be the human-AI blends Ahrefs identified — drafted fast, then revised hard. Raw output competes; edited output ranks.
Do readers trust AI-written content?
Less than marketers might hope, and trust is the binding constraint. A Gartner survey of 377 US consumers found 53% distrust AI-powered search results, 61% want a toggle to turn AI summaries off, and 41% say AI overviews make search more frustrating.
The reaction to content is just as pointed. Bynder surveyed 2,000 US and UK consumers and found 52% become less engaged once they suspect copy is AI-generated. The twist: in a blind test, 56% actually preferred the AI-written version — so the penalty is triggered by disclosure and suspicion, not always by quality.
Are AI content detectors reliable?
No, and the false-positive risk falls hardest on real writers. A widely cited study in Patterns by Liang et al. ran seven commercial GPT detectors against essays by non-native English speakers and flagged more than 61% as AI-generated, while classifying native-speaker essays with near-perfect accuracy. At least one detector flagged 97.8% of the non-native essays.
The mechanism matters: detectors keyed on low “perplexity,” and simpler, more predictable prose reads as machine-like whether a person or a model wrote it. Prompting a rewrite to sound more native dropped the misclassification rate from 61.3% to 11.6% — evidence that detectors measure style, not authorship.
Frequently asked questions
What percentage of content marketers use AI?
Eighty-seven percent use AI to create or help create content, and only 13% do not, per a survey of 879 practitioners by Ahrefs. Semrush puts the share drafting copy with AI at 76%. Across surveys, AI writing is now the default rather than the exception for marketing teams.
How much of the internet is AI-generated?
Ahrefs scanned 900,000 new web pages in April 2025 and found 74.2% contained AI-generated text. But only 2.5% were pure AI. The dominant form is a human-AI blend at 71.7%, with 25.8% still purely human-written — so “AI-assisted” describes the web far better than “AI-written.”
Does AI-generated content rank in Google?
It ranks, but rarely at the top. Semrush analyzed 42,000 blog pages and found human-written content holds an 80.5% probability of position one versus 10% for AI text, keeping an edge across all top-10 positions. Lightly edited AI content competes; edited, human-led content wins the best slots.
How much time does AI save writers?
Meaningful time on first drafts. Per Semrush, 36% of AI users finish a long-form post in under an hour, while 38% of non-users take 2-3 hours. HubSpot reports 74% of marketers believe AI improves their productivity. The savings concentrate in drafting, not in research or editing.
Do consumers trust AI-written content?
Not reliably. A Gartner survey found 53% distrust AI-powered search results, and Bynder found 52% disengage once they suspect copy is AI-written. Notably, 56% preferred the AI version in a blind test — so the trust penalty is triggered by disclosure and suspicion, not by quality alone.
Can AI detectors reliably catch AI writing?
No. Research in Patterns by Liang et al. showed seven commercial detectors flagged over 61% of non-native English essays as AI-generated while nearly always clearing native speakers. Detectors measure predictable, low-perplexity style rather than true authorship, which makes them unsafe as a basis for penalizing writers.
What this means for marketers
The data draws a clean line for the 80/20 buyer. AI writing tools save real time on drafting — enough that 87% of teams adopted them — but readers still disengage the moment content feels machine-made, and detectors are too unreliable to police it. The value is in the human-AI blend, not in automation. That means one capable AI writing tool plugged into a disciplined editing process beats a stack of six overlapping generators.
Pick the tool that drafts fastest in your voice and integrates with where you already publish, then spend the saved hours on the editing that makes AI output rank and earn trust. The same logic runs through our marketing automation statistics: fewer, better-chosen tools beat a longer shortlist. See how we score each pick on the about page.
Sources
- Ahrefs — What Percentage of New Content Is AI-Generated? (Study of 900k Pages)
- HubSpot — AI in Content Marketing [Data]
- Semrush — 96 Content Marketing Statistics You Need to Know
- Semrush — Does AI Content Rank Well in Search? [Survey + Data Study]
- Gartner — 53% of Consumers Distrust AI-Powered Search Results
- Bynder — AI vs Human-Made Content Study
- Liang et al., Patterns — GPT Detectors Are Biased Against Non-Native English Writers