The article referenced above, authored by Francesco Agnellini from Binghamton University and published yesterday, cites a recent study conducted by Graphite. This study analyzed over 65,000 randomly selected articles and reports that, as of 2025, more than 50% of newly published English-language web content is generated by AI. Notably, as early as 2022, the European Union projected that as much as 90% of online content may be synthetically generated by 2026, indicating that this prediction was remarkably prescient https://op.europa.eu/en/publication-detail/-/publication/b844aa2b-dbd4-11ec-a534-01aa75ed71a1/language-en
It is crucial to emphasize that the Graphite study may underrepresent the true prevalence of AI-generated content online. This underestimation stems both from the relatively small sample size and from limitations in detection tools, which can misclassify human-authored articles assisted by AI as purely human-written. Let us not forget that, just nine months ago, Dirk Spennemann, in his study, estimated that 30–40% of online content is AI-generated, highlighting the rapid proliferation of synthetic text. Taken together, these observations suggest that the actual share of AI-influenced content may be considerably higher than reported.
Agnellini’s article is significant because it highlights a paradox at the heart of today’s digital landscape: while AI increasingly saturates the web with unprecedented volumes of text, what may become rarer—and consequently more valuable—are the authentic human voice, genuine originality, and intentional stylistic expression. His reflections rightly underscore the dangers of eroding creativity, diversity, and the singularity of human expression. Yet they neglect a more alarming danger: the spiraling loop of AI-generated content feeding upon itself. This self-referential drift raises a critical question: if what we call “originality” becomes increasingly absorbed into algorithmic patterns and human creativity slowly recedes into the background, must we not ask whether the internet will continue to embody the human condition—or whether it is evolving into a reflective surface that reveals nothing but machines interpreting themselves?
