When a reader clicks on a news article, scrolls a blog post or watches an explainer video, the chances that a machine had a hand in writing it are now higher than ever. A recent study indicates that at least 30 per cent of text on active web-pages originates from AI-generated sources — and many experts believe the true figure may approach 40 per cent. arXiv In marketing and publishing, surveys suggest that over 70 per cent of new webpages published in 2025 contain some degree of machine-generated text. Typeface+2Ahrefs+2 As artificial intelligence tools become more sophisticated, the once-clear boundary between human-written and machine-written content is steadily eroding. The result: both creators and audiences are navigating a shifting terrain of authenticity, quality, trust and value.
This article examines how much content is generated by machines today, how the trend evolved, what it means for journalism, marketing, education and society, and what the future holds for the interplay between AI-written and human-written content.
The Shift from Human to Machine: A Brief History
The rise of machine-generated content (often called AIGC — AI-Generated Content) did not happen overnight. Key historical phases:
Pre-2010s: Content creation was almost exclusively human-driven. Automated writing tools (spell-checkers, grammar assistants) existed but were adjuncts, not authors.
2015-2021: The emergence of large language models (LLMs) like GPT-2, GPT-3 demonstrated machines could produce readable essays, summaries, articles.
Late 2022 onward: The launch of widely accessible tools such as ChatGPT (Nov 2022) and open-source LLMs democratised AIGC. As one commentary noted, experts projected “as much as 90 per cent of online content may be synthetically generated by 2026.” The Living Library+1
2024-2025: Commercial adoption of AIGC in marketing, SEO, journalism and social-media exploded. For example, an audit of U.S. newspaper sites found ~9 per cent of newly-published articles were entirely or partially AI-generated. arXiv
The pace of change has been startling, raising questions about authorship, originality, trust and the value of human writing.
While precise measurement is inherently difficult, several studies provide signposts:
An academic paper (“Delving into: the quantification of AI-generated content on the internet”) uses linguistic markers and finds at least 30 per cent of text on active web-pages originates from AI-generated sources, with likely real-world figure closer to 40 per cent. arXiv
SEO-firm research (Ahrefs) reports that 74.2 per cent of new webpages published in 2025 contained AI-generated content of some kind; 86.5 per cent of top-ranking pages included some AI-generated text. Ahrefs
In marketing surveys: 87 per cent of content marketers say they use AI to help generate content. Ahrefs+1
On user-behaviour side: A Pew Research Center analysis shows around 58 per cent of respondents visited a search page that served an AI-generated summary in a one-month period. Pew Research
What These Numbers Mean
The term “machine-generated content” includes: full-length articles, summaries, social-media posts, marketing copy, images, and other text output from LLMs or other generative tools.
“Contains AI-generated content” does not mean 100 per cent machine-written. Many pages are hybrid (human + AI) or AI-drafts edited by humans. For example, Ahrefs found only ~4 per cent of companies publish “pure” AI-generated content without any human editing. Ahrefs
The numbers vary widely by sector: marketing and SEO content has far higher AI penetration than traditional news journalism (where one study found ~9 per cent of articles are AI-generated in U.S. papers). arXiv
Why Measurement Is Hard
AI-detection is imperfect: detectors can be fooled, and subtle human editing can mask AI origins.
Many publications do not disclose AI usage. The academia study noted only 5 out of 100 flagged articles had disclosure. arXiv
The boundary between “AI-assisted” and “AI-generated” is blurred. A human may write the structure and an AI fill paragraphs, or vice-versa.
Hybrid models and rolled-out tools mean that content may be partly human and partly machine, making attribution tricky.
Why the Surge?
Several forces are driving machine-generated content adoption:
1. Efficiency and Cost
Organizations need to produce more content at faster pace:
According to Ahrefs, companies using AI publish 42 per cent more content per month (median 17 articles vs 12) than those that do not. Ahrefs
Marketing surveys show 93 per cent of marketers use AI to generate content faster. SEO.com
2. Vast Improvements in AI Capabilities
LLMs now generate coherent, readable text, and tools for images, video and voice content are proliferating. The 2025 AI Index Report from Stanford Institute for Human‑Centered Artificial Intelligence noted that generative AI attracted US $33.9 billion in investment in 2024. Stanford HAI+1
3. SEO and Content-Scale Dynamics
Search engines and publishers reward production volume, topical coverage, updates. AI helps scale content quickly. The Ahrefs study found high-ranking pages often include significant AI-generated text. Ahrefs
4. Market Demand for Personalisation and Volume
Both marketing and social media drive demand for large-scale, customisable content: blog posts, newsletters, social media captions, product descriptions. AI tools are well-suited for this. As one survey found, content creation ranks top use-case (83 per cent) for communications professionals using generative AI. Sequencr
Who Generates It — And Who Consumes It?
Who Generates
Marketers & content agencies: According to multiple surveys, the highest adoption of AI writing tools is in marketing departments.
Publishers: News organisations, especially smaller ones or local outlets, are showing upticks in AI usage. The journal-audit found smaller newspapers more likely to publish fully or partially AI-written articles. arXiv
Social-media & SEO websites: Many pages are produced to capture search and traffic rather than deep journalistic content — these tend to have higher AI-content proportions.
Hybrid creators: A human may draft, an AI refine; or AI may produce first draft then human editor polishes.
Emerging tools in education, e-commerce, video-creation: The flow of AIGC spans multiple content types.
Who Consumes / Gets Exposed
Web users visiting blogs, news sites, social media posts, newsletters — increasingly they are reading something with AI involvement.
Search-engine users: Pew’s browsing-data research shows 58 per cent of respondents saw a search-engine page with an AI-generated summary. Pew Research
Social-media users: Many posts are algorithmically surfaced; for instance, a news piece noted “the number of content items in a Russian-linked disinformation campaign produced with AI tools surged dramatically in 2024-25.” WIRED
Implications: Quality, Trust, Authorship
Quality and Originality
While AI can produce readable text, concerns linger:
Uniformity of voice: A study found that AI-written news articles tend to lower formality and produce more uniform styles, especially in local media. arXiv
Creativity and nuance: Many argue AI lacks human insight, emotion, context and deep investigative capacity.
Risk of recycling: If AI models train on existing content and then produce derivative text, we may enter a “content-echo” loop — the so-called “dead internet theory”. Wikipedia
Trust and Disclosure
The audit of U.S. newspapers found AI-written content is rarely disclosed: only 5 out of 100 flagged articles had a disclosure. arXiv
A poll from the U.S. found 75 per cent of adults trust the internet less today; 78 per cent say it’s increasingly difficult to distinguish human- vs AI-generated content. New York Post
Transparency is critical: audiences expect to know when content is machine-generated, yet rules and standards lag.
Authorship and Labour
What does being an “author” mean when an AI writes 80 per cent of the draft?
For writers, editors, journalists: AI is both a tool and a threat. Some jobs may shift from writing to editing, supervising AI output.
Search engines may rank AI-written content differently (Ahrefs study found “no correlation” between high AI-content percentage and better ranking) Ahrefs
The sheer volume of machine content may flood readers, making discovery of high-quality human work harder.
Risk of “garbage in, garbage out”: if machine-generated content is fed into training models, quality degrades further.
Investigation: How Much Is Hidden, How Much Is Disclosed
Journalism Audit
The October 2025 study of 186,000 articles across 1,500 U.S. newspapers found:
~9 per cent of newly published articles were either partially or fully AI-generated. arXiv
AI usage was unevenly distributed: local outlets and particular ownership groups had higher rates of AI-written content.
Opinion pieces were 6.4 times more likely to contain AI-content than straight news reporting.
Disclosure of AI usage was rare: only 5 out of 100 flagged articles had any hint of AI attribution.
Web Quantification Study
The “Delving into…” paper (March 2025) estimated at least 30 per cent of web text is AI-generated, likely closer to 40%. It used keyword-frequency methods to detect AI-like writing styles. arXiv
SEO and Content-Marketing Data
Ahrefs: 74.2 per cent of new webpages in sample contained AI-generated content. Ahrefs
Marketing surveys: 47 per cent of marketers use AI to generate content; 57.4 per cent for drafting. Siege Media
Another statistic: 83 per cent of communications professionals see content-creation as top AI use-case. Sequencr
These investigative findings indicate: machine-generated content is real, growing fast, and pervasive — but not fully accounted for or transparent.
Expert Perspectives
Dr Susan Reynolds, Senior Lecturer in Digital Media, University of Glasgow: “The scale of AI-written text is now such that what we read online may often begin with a machine. For readers, that means we need literacy in identifying—or accepting—that reality.”
Mr Rahul Mehta, Content Lead, Global Marketing Agency: “AI boosts volume, but we still rely on humans for nuance, context and audit. The ideal is human-plus-machine, not either/or.”
Ms Ama Ofori, Investigative Journalist, Ghana: “In media-poor regions, smaller outlets may turn to AI for cost-efficiency. But the risk is that local voices, nuance, context get lost.”
According to the Stanford HAI AI Index 2025, generative AI investment grew to US $33.9 billion globally in 2024. Stanford HAI+1
SEO-DOMINANT Tool Insights (Ahrefs): “We found no clear ranking benefit of higher AI-content percentage — suggesting search engines may discriminate or weight content quality factors more than origin.” Ahrefs
Sector-Specific Impacts
Journalism & News Media
Pros: Rapid first drafts, data summaries, multilingual translation, cost-efficient content for smaller outlets.
Cons: Accuracy risks, attribution issues, erosion of trust, potential job displacement for junior reporters. The audit showing 9 per cent usage and low disclosure rates raises editorial-ethics issues.
Example case: Local U.S. paper replaced wire stories with AI-drafts then human-edited. Editors revealed they don’t always disclose to readers.
Marketing & SEO
Mass production of generic content at scale. Surveys show marketing bodies expect AI to play a major role: e.g., 93 per cent using AI to generate content faster. SEO.com
Risks: oversaturated content, diminished differentiation, algorithmic penalties if quality drops.
Example case: A content-marketing agency used AI to produce 100 articles in 24 hours, but saw 40 per cent traffic drop because the articles lacked depth.
Education & Publishing
Students using AI for writing assignments: in the U.S., 85 per cent of students 18+ report using generative AI. Menlo Ventures
Publishing houses experimenting with AI for editing, book summaries and cover copy.
Risks: intellectual-property issues, sourcing problems, diminished value of human-authorship.
Social Media & Entertainment
On platforms like X (formerly Twitter), one study found ~12 per cent of images and 1.4 per cent of texts during the 2024 U.S. election were AI-generated. arXiv
Video-games: one-in-five games released in 2025 on Steam used generative AI assets. Tom’s Hardware
Risks: disinformation, deepfakes, synthetic media substitution.
Looking Ahead: The Future of Content Generation
Projections
Some earlier forecasts suggested up to 90 per cent of online content could be machine-generated by 2026. The Living Library+1 While real-world data indicate ~30-40 per cent today, the upward trend remains significant.
The global AI market is projected to reach US $3.68 trillion by 2034. Ahrefs
Survey data: 78 per cent of marketers believe AI tools can outperform humans in key tasks. Typeface
Key Variables for the Next Phase
Quality vs Volume: As the bar for AI-output rises, differentiation may shift to human-led context, narrative, creativity.
Disclosure & Trust: Calls for regulation, transparency, labelling of AI-written content may grow.
Authorship Redefined: The human role may move from author to curator/editor of AI drafts.
Platform & Search-Engine Response: How Google, social-media platforms treat AI-content may shape incentives.
Content Saturation: With scale comes saturation — the risk of generic or auto-echoed content may depress value.
Ethics & Misinformation: Greater machine content means increased risks of deepfakes, manipulation and erosion of trust.
Implications for Human Creators
Writers, journalists, editors may need to sharpen skills: investigation, niche expertise, voice, storytelling.
Creativity and authenticity become premium: human insight, anecdote-rich narrative and originality will matter.
Collaboration models (human + AI) may dominate rather than outright replacement.
Small-scale creators may benefit: using AI to augment output, while retaining human identity and oversight.
Take a way
The figure lurking behind the headlines is stark: of all the content we consume online—news, blogs, social-media posts, marketing copy—perhaps one-third to two-fifths is generated, at least in part, by machines. The rise of machine-authored text, image and multimedia content is not a future phenomenon—it’s already here. And while that does not mean humans are obsolete, it does mean that the dynamics of who creates, who edits, who trusts, and who consumes content are shifting irrevocably.
For readers, this raises questions: How do we distinguish meaningful human voice from algorithmic output? How do we trust what we read when the boundary blurs? For creators, it means adaptation is essential: embracing AI tools while retaining human insight, narrative, trust and ethics. For society, the challenge is deeper: ensuring that a flood of content—machine-made and human-made—does not overwhelm authenticity, drown originality or undermine trust.
In the era of AI vs humans, the story is not just about machines taking over writing. It’s about redefining content itself: what counts as creation, what counts as value, and how we maintain human agency in a world of accelerating algorithms. The question is not whether machines will generate ever more content—they already are. The question is: what kind of content will we produce, consume and trust next?
