GoKawiil
Tech News
A Duke journalism professor and PolitiFact founder, Bill Adair, found that Grokipedia, Elon Musk's AI-written encyclopedia launched a year ago, confused his biography with that of a different, deceased man of the same name, inventing a baseball career and other false details. Despite the errors and bloated writing, Adair notes Grokipedia surfaces obscure details from across the internet that Wikipedia often omits.
A Show HN author published a personal essay revealing they are a convicted felon, something they had kept hidden for two decades. In the same post they explain their writing process, saying they compose text by hand in Neovim without AI assistance, while using AI tools extensively for coding work that they still personally review.
An author drawing on their bachelor's thesis on stylometric authorship detection argues that AI-written text 'averages style,' making it difficult to identify a distinct author or empathize with one. They contend AI text also loses precision because models must make assumptions to stitch sentences together, often choosing vaguer language, and because writers skip the self-challenging process that comes from manually drafting ideas. The result, they say, is text that is slower to read despite being the same length as human writing.
An independent developer released qwen3-4b-deslop, a fine-tuned version of Qwen3-4B-Instruct trained on a MacBook Air using 161 example answers to remove stylistic habits like em dashes, 'not X, but Y' constructions, and lists of three. The model is available as a 2.5GB download runnable via Ollama, and on 40 held-out test prompts its measured 'slop score' fell from 121 to 12.5 per 1000 words while maintaining accuracy on 30 factual questions.
A new study examining Claude Opus 5.5 found the model frequently relies on certain words and phrases, including 'dependable' and 'quietly', that can signal AI-generated text. The research reportedly found that em dashes, long considered a giveaway of AI writing, are not a reliable marker for this model.
Marketing firm Graphite analyzed writing from frontier AI models by having them rewrite 10,000 pre-ChatGPT articles from summaries, then compared the results to the original human text. The study identified over 13,000 words and phrases that appear at least twice as often in AI writing as in human writing, with each model showing distinct habits—Claude Opus 5.5, for example, overuses 'dependable' and phrases like 'this matters.'
Dartmouth College is investigating provost Santiago Schnell after its student newspaper, The Dartmouth, found signs of AI-generated text in writing published under his name. Schnell said he used AI for editing and clarity, but when the newspaper ran his own explanatory statement through an AI detector, it also flagged that text as AI-generated. The controversy began after Semafor reported on AI detection tool Pangram finding AI use across opinion pages at major outlets.
A recent controversy in France's literary world has intensified public scrutiny over whether written and spoken content is AI-generated. Commentary describes attending a conference where a speaker's talk carried recognizable ChatGPT phrasing, such as stock transition lines, prompting the observer to disengage from the presentation.
Researchers Liu, Lee and Bai ran a large between-subjects experiment where participants wrote hiring evaluations of resumes from a 'male' and 'female' applicant using an LLM autocomplete tool. When suggestions countered gender stereotypes, the female candidate's perceived competence and salary offers rose to match the male candidate's, but she was rated less likable, and the male candidate remained the overall favorite. Participants largely didn't notice the stereotype manipulation, focusing instead on the writing tool's usefulness.
Benchmarking site Arena analyzed high-reasoning Text Arena responses from August and September 2026 and found Anthropic's Claude Opus 5.5 uses far fewer em dashes and semicolons than Opus 5, along with shorter, simpler sentences. Em dash usage fell from 15.2 to 0.8 per 1,000 words, and average sentence length dropped from 12.14 to 10.03 words, though overall answers grew longer, from 453 to 481 words on average.
A software professional describes a growing trend where colleagues use AI to retroactively generate design documents, pull request summaries, tickets, and even personal messages after work is already done. The writer argues these AI outputs are technically detailed but lack context, perspective, and human judgment, making them exhausting and frustrating to read despite AI's usefulness for improving writing speed and quality elsewhere.
A tech commentator argues that while AI tools like Claude, Codex/ChatGPT and Gemini are useful for coding and research, they should not be used to generate written documents from scratch. The reasoning is that writing itself is the process of thinking through a problem, and outsourcing that to AI means skipping the deep thinking required to produce genuinely useful, well-considered content. The author does allow for AI to be used afterward, to polish writing once the original thinking and drafting is complete.