{"id":227959,"date":"2026-05-31T14:05:32","date_gmt":"2026-05-31T14:05:32","guid":{"rendered":"https:\/\/teknomers.com\/en\/ai-is-already-winning-literary-awards-the-only-way-we-can-prove-our-humanity-is-by-writing-poorly\/"},"modified":"2026-05-31T14:05:34","modified_gmt":"2026-05-31T14:05:34","slug":"ai-is-already-winning-literary-awards-the-only-way-we-can-prove-our-humanity-is-by-writing-poorly","status":"publish","type":"post","link":"https:\/\/teknomers.com\/en\/ai-is-already-winning-literary-awards-the-only-way-we-can-prove-our-humanity-is-by-writing-poorly\/","title":{"rendered":"AI is Already Winning Literary Awards: The Only Way We Can Prove Our Humanity is by Writing Poorly"},"content":{"rendered":"\n
\n

On May 18, social media lit up when a group of users on X began to vocally express their doubts regarding the latest Commonwealth Prize winner in the short story category. Published in ‘Granta’, a highly regarded British literary magazine, the winning narrative was suspected to be generated by ChatGPT. To complicate matters, the author’s photo appeared suspiciously artificial as well. When confronted, the magazine\u2019s vague responses only muddled the situation further. They posed a question to Claude, the AI model, inquiring if the text sourced from AI, to which it responded in the negative.<\/p>\n

<\/p>\n

How AI Text is Detected<\/h2>\n

Distinguishing AI-generated writing is more complex than one might think. Identifying the prose of a language model becomes feasible once you\u2019ve trained your eye to recognize certain patterns. Language models don’t aim for precise word selection; instead, they generate the statistically most likely next word based on context. This method reveals identifiable traits. For instance, the common phrase structure “it\u2019s not X, it\u2019s Y” serves as a sort of Rosetta Stone for identifying AI-generated text. Furthermore, excessive metaphors and underwhelming verb choices like “delve into” are prevalent. This mechanical replication of stylistic tropes makes the AI’s writing often too formulaic.<\/p>\n

Identifying Weaknesses in ‘Granta’<\/h2>\n

In ‘Granta’s winning story, evocative phrases like “the humming noon” and “sweet air with the smell of cane and oblivion” can be misleading. Analysts, including Benjamin Breen, have pointed out that such expressions convey an attraction to vague emotional states, like nostalgia and sadness, which AI models struggle to genuinely understand. Therefore, when a language model indiscriminately deploys sensory details without real nuance, it becomes an easy target for those skilled in recognition.<\/p>\n

<\/p>\n

\n
\n
\n
<\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n

Why Detection Tools Aren’t Reliable Yet<\/h2>\n

Recognizing intuitive writing is one challenge, yet proving it objectively is another. Tools like GPTZero and Originality.ai have strongly incorrect identification rates. OpenAI even phased out its AI Text Classifier in July 2023 because it accurately identified only 26% of AI-generated text and misidentified nearly 9% of genuine human work as AI-produced.<\/p>\n

<\/p>\n

The lone exception noted is Pangram, which utilizes a technique known as mirror data<\/em>. This trains classifiers with pairs of stylistically similar texts from different sources, achieving false positive rates nearing zero and false negatives between 2% and 4%. However, skepticism remains surrounding any tool that markets a “humanizer,” as these often present unreasonably inflated figures.<\/p>\n

<\/p>\n

Educational Implications<\/h2>\n

While the ‘Granta’ scandal sheds light on challenges within the publishing world, a parallel crisis unfolds in academic institutions. A recent report highlighted a cycle where teachers regularly run student assignments through AI detection tools. Misidentifications encourage students to either resort to overpriced humanizers or even intentionally write in a less sophisticated manner to avoid being flagged.<\/p>\n

Joseph Thibault from Cursive has documented the use of 43 humanizers with over 33.9 million views combined. Tools like Grammarly\u2019s Authorship have also emerged, allowing students to record their writing processes as proof of authorship. However, many educators have revealed that strong writing often leads AI detectors to mistakenly tag them as machine-generated.<\/p>\n

\n
\n
\n
\n \"The\n <\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n

Wider Industry Trends<\/h2>\n

The ongoing situation mirrors a broader trend within the publishing domain. Recent analytics reveal a significant disparity between non-fiction and fiction sales. For instance, in April, the non-fiction bestseller sold only 13,468 copies in its debut week, while the leading novel shifted nearly 105,000 copies. This shift is attributed to the rise of podcasts, which engage listeners more swiftly than traditional texts, echoing patterns observed with platforms like YouTube.<\/p>\n

AI tools are not merely competitors but could replace traditional literature altogether, providing instantaneous summaries that bypass the need for human input altogether. The challenge, as noted by Breen, lies in the abandonment of depth and reflective nuance\u2014qualities that nonfiction books inherently offer.<\/p>\n

The Impact of AI on Reader Preferences<\/h2>\n

In a thought-provoking experiment, author Vauhini Vara used AI to craft text resembling his own work, blending it with his own writing before flying under the radar of even his closest friends. The surprising find? Readers often preferred AI-generated text until they learned its origin. This underscores a deep-seated truth: what readers truly crave is human authenticity attached to the words they consume.<\/p>\n<\/div>\n


\n
General News – 2<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

On May 18, social media lit up when a group of users on X began to vocally express their doubts regarding the latest Commonwealth Prize winner in the short story category. Published in ‘Granta’, a highly regarded British literary magazine, the winning narrative was suspected to be generated by ChatGPT. To complicate matters, the author’s […]<\/p>\n","protected":false},"author":1,"featured_media":227960,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[36399],"tags":[11314,6796,3657,4394,7718,1650,13930],"class_list":["post-227959","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology","tag-awards","tag-humanity","tag-literary","tag-poorly","tag-prove","tag-winning","tag-writing"],"_links":{"self":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts\/227959","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/comments?post=227959"}],"version-history":[{"count":1,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts\/227959\/revisions"}],"predecessor-version":[{"id":227961,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts\/227959\/revisions\/227961"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/media\/227960"}],"wp:attachment":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/media?parent=227959"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/categories?post=227959"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/tags?post=227959"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}