ai writing

Bilingual

The End of an Era
2 months ago
- The author reflects on a lucky 20-year window (roughly 2007-2024) when writing was hard but publishing became easy, which is now closing due to AI.
- An example of a $2.4M advance for a debut novel was rescinded after AI-use concerns, illustrating how every book is now suspect.
- AI-generated books are already winning awards, crowding bestseller lists, and potentially getting huge advances, signaling an end of an era.
- Prediction: Publishers will eventually embrace AI books, editing them to hide AI fingerprints, leading to machine-books sold alongside human-written ones.
- Most readers won't care whether a book is written by a human or AI, just if it's good, similar to how readers ignored self-publishing stigma.
- A niche group of readers will specifically seek out machine-books, similar to chess enthusiasts who admire specific AI engines.
- Established authors will increasingly use AI for assistance in research, editing, marketing, and other tasks, changing the writing process.
- The author's heart breaks for artists facing confusion and anger as it becomes easier to fake being an artist, making it harder for genuine artists to earn a living.
- There will be 'meat-book fanatics' who value human authors and their stories, leading to tools that prove a book's human origins, possibly through livestreaming or blockchain.
- The author advises doing creative work for love, not for external rewards, and welcomes readers to the new era, which is less favorable than the past decade.
Which Design Doc Did a Human Write?
2 months ago
- The author created three design docs for the same open-source web app: one handwritten (16 hours), one by Claude Opus 4.6 (medium effort), and one by GPT-5.4 (high effort), then asked readers to identify the human version.
- Readers correctly identified the human-written doc by its personal opinions, anecdotal experience, unusual technology choices (e.g., NixCI, PolyForm-Noncommercial license), and non-generic architecture headings.
- AI-generated docs were spotted by 'bloat'—verbose, fluffy sentences, overuse of bold text, unrealistic time estimates (like '10 hours' milestones), and generic security/privacy sections lacking app-specific threats.
- Images were a dead giveaway: AI-generated diagrams had layout problems with colliding elements, so the author stripped diagrams from all versions.
- Both AI versions missed the hard problems—such as balancing security and usability in a photo-sharing app—and focused on generic best practices instead of the specific challenges of the project.
- The author concluded that human-written design docs convey the tough design trade-offs and competing interests, while AI tends to produce generic, bloated content.
The Empty Promise of AI-Generated Creativity
2 months ago
- Yarn Spinner will never integrate AI that writes for you, based on consistent disappointment with AI-generated narrative content.
- AI-generated stories are derivative, hollow, and lack coherent creative threads, relying on mimicry rather than genuine creation.
- AI lacks lived experience, cultural understanding, and authentic purpose, which are essential for meaningful creative work.
- Ethical concerns include AI systems trained on human works without permission, appropriating creative labor.
- Yarn Spinner aims to amplify human creativity, not replace it, and opposes normalizing AI in creative processes.
- Human creativity, with its unique perspective and intention, remains irreplaceable and worth protecting against technological trends.
Human Routers of Machine Words
2 months ago
- Reading AI-generated content feels like a trick, evoking offense, sadness, and contempt for authors who use AI to write.
- Defending AI-assisted writing by claiming 'ideas are mine, writing is AI's' is flawed because ideas and writing are inseparable.
- Ideas are vague and nebulous; writing is the process that concretizes and refines them, revealing flaws and contradictions.
- Writing is the ultimate form of thinking, not a mere transcription of pre-formed ideas.
- AI-generated text shifts the burden of critical thinking to readers, who must verify coherence.
- The issue is not AI capability but the dishonesty and irresponsibility of humans who rely on AI to produce generic, unoriginal output.
Will the explainer post go extinct?
2 months ago
- Will short-form non-fiction internet writing go extinct?
- AI can replicate explanations with customized styles and answer questions indefinitely, threatening human-written explainers.
- Human-written explainers currently offer advantages like accuracy, style, parasocial relationships, trust, and social value.
- AI accuracy is key: if AI matches human accuracy, explainer posts may become obsolete.
- Potential counterarguments include AI improving content discovery, new reward systems, and human-AI collaboration.
The modern formatting addiction in writing
2 months ago
- The author contrasts two writing styles: Exhibit A (heavily formatted with bullet points and sections) and Exhibit B (traditional paragraphs).
- Excessive formatting is disorienting and violates the principle of making something you would actually enjoy reading.
- Both humans and AI models exhibit a similar addiction to formatting, even when it hinders clarity.
- Formatting may be useful for skimming and SEO content, but it is excruciating for long-form essays.
- Formatting aids in thinking and structure during the writing process but often remains unrefined.
- Formatting can bluff coherence, concealing a lack of synthesis and preventing deeper narrative development.
- Writing in paragraphs forces confronting contradictions and improves clarity; the author recommends starting with formatting then transitioning to paragraphs.
Blink if you’re human
2 months ago
- The author guarantees all words are typed manually, with exceptions for quotes and labeled AI outputs.
- The guarantee serves as reputational risk because future AI detectors may catch deception.
- Using AI for writing undermines proof of effort and parasocial connection in blogs.
- AI-generated content risks creating a 'lemon market' where readers distrust all essays.
- The author suggests writers who restrict AI use should state their limits publicly rather than forcing universal disclosure.
- AI use exists on a spectrum from minimal assistance to full generation, and most writers should find a personal boundary.
- The author self-assesses at about 10% on the AI usage spectrum, avoiding direct copying but using AI for research and minor edits.
- Drawing a clear line for oneself helps prevent gradual drift into heavy AI dependence.
Have you read it?
2 months ago
- AI-written articles often lack depth because the AI 'author' is discovering the content simultaneously with the reader.
- A good insight can be compelling despite stylistic flaws, but the absence of genuine human thought behind it undermines its value.
- Readers cannot question the author about the article's insights, as the AI did not consciously think or intend them.
- The core problem is that the article has no real author behind it, making it feel hollow and unrelatable.
The whole premise of checking for human writing is daft
2 months ago
- Typing every word by hand is an outdated test of human thought; intelligence lies in intention, judgment, and responsibility, not in physical keystrokes.
- A person can type thousands of words mindlessly, while another can use AI to refine ideas and produce thoughtful work, yet current rules unfairly label the latter as suspicious.
- Historical tools like fire, calculators, and spellcheck were initially criticized for diminishing human abilities, but eventually became ordinary and expanded human potential.
- The demand for proof of human writing excludes people with disabilities who rely on assistive technology, such as Stephen Hawking or users of mouth-controlled devices.
- Authorship depends on who owned the ideas, made the decisions, and takes responsibility for the work, not on how words were physically produced.
- Schools and employers can test specific skills when needed, but making unaided typing a universal test of humanity is technologically backward and exclusionary.
AI Slop: Why Philosophy Journals Should Reject AI-Written Prose
2 months ago
- Human-generated text reflects authentic expert thought and subtle word choices that carry philosophical significance, unlike AI-generated approximations.
- Knowing that a text is human-generated provides meta-epistemic evidence of its quality, similar to the value of journal imprimatur.
- AI can be used for copyediting only if done actively (e.g., typing revisions manually) to preserve rather than blur the writer's distinctive phrasing.
- The argument is bounded: if AI surpasses human expertise in word choice, it may no longer apply, but for now, human prose is preferable.
AI-isms go deeper than em-dashes and 'load-bearing'
2 months ago
- AI writing often overuses em-dashes and alien vocabulary like 'load-bearing', 'gated', and 'belt-and-braces'.
- AI frequently attributes agency to powerless objects, e.g., saying a lateral join 'rides' an index or code 'moved out'.
- Humans would not describe code hunks as 'clear blends' or use such agentic language for inanimate things.
- The author hypothesizes AI may be pushing 'ontological egalitarianism', treating all objects as agentic.
- These tics stem from AI training emphasizing active voice, which is inappropriate for technical descriptions.
The Most Famous AI Writing Tic Is Also the Most Mysterious
2 months ago
- AI writing frequently uses the rhetorical device 'It's not X, it's Y' or negative parallelism, making it a telltale sign.
- This construction has significantly increased in corporate communications and AI-generated text, with chatbots like ChatGPT and Claude heavily relying on it.
- Experts theorize AI favors negative parallelism due to reinforcement learning from human reviewers and its role as a safe, predictable sentence structure.
- Fixing the issue is challenging due to AI models training on their own output, creating a feedback loop that perpetuates the device.
- While annoying, AI clichés help distinguish machine writing from human, though they risk making a once-effective rhetorical tool seem like a bot trademark.
Claude's writing style has me on edge
3 months ago
- Benn Stancil's column discusses AI writing, particularly a paragraph from Claude that seemed profound but was empty nonsense.
- The author initially thought Benn might have outsourced writing to Claude, but then learned it was AI-generated text.
- The experience reveals the author's unease with how prevalent Claude's writing style has become across platforms like Reddit, X, Hacker News, and LinkedIn.
- The author feels exhausted by constantly trying to discern if content is human or AI-written, adding a meta layer to reading.
- The piece reflects on Claude's evolving voice and the blurring line between human and AI-generated text, prompting broader concerns about authenticity in writing.

ai writing

The End of an Era

Which Design Doc Did a Human Write?

The Empty Promise of AI-Generated Creativity

Human Routers of Machine Words

Will the explainer post go extinct?

The modern formatting addiction in writing

Blink if you’re human

Have you read it?

The whole premise of checking for human writing is daft

AI Slop: Why Philosophy Journals Should Reject AI-Written Prose

AI-isms go deeper than em-dashes and 'load-bearing'

The Most Famous AI Writing Tic Is Also the Most Mysterious

Claude's writing style has me on edge