ai-generated content

Bilingual

AI broke writing as proof-of-thought
2 months ago
- AI has broken the traditional link between fluent writing and deep understanding, making it possible to produce convincing prose without genuine thought.
- Readers have an innate fluency bias that makes them trust smooth writing, even when it's AI-generated, and awareness alone does not neutralize this bias.
- AI-generated content creates a false feeling of understanding in readers, who mistake surface-level fluency for true comprehension.
- The most dangerous users of AI are those with some domain knowledge who believe AI output reflects their own understanding, leading them to publish pseudo-knowledge.
- Writing without AI is essential for genuine understanding, as the process of explanation reveals gaps in thinking that prompting does not.
- The author advocates for provenance in writing and supports AI detection tools to protect the signal of human thought in published content.
LinkedIn and X Are Flooded with AI Spam, Browsing Data Suggests
3 months ago
- Pangram data shows that 41% of longform LinkedIn content and about one-third of longer X posts are likely fully AI-generated.
- The study suggests AI content is prevalent on popular platforms like Reddit and Substack, with longform posts being more commonly AI-generated than shortform.
- AI-generated content is not confined to obscure sites; users regularly encounter it on major social media platforms during normal browsing.
- The research used a Chrome extension to passively scan posts users actually see, providing insights into the real-world exposure to AI content.
- Platforms like LinkedIn have built-in AI tools, but are now trying to disincentivize AI content, while Reddit moderates against AI spam.
- AI detectors like Pangram are not foolproof, but the data likely underestimates the problem, as users who install such extensions may avoid AI content.
AI Content Is Everywhere on Social Media, Especially LinkedIn
3 months ago
- Pangram's Chrome extension scans social media posts to flag AI-generated content, helping users make informed decisions about their attention.
- Data from opt-in users shows an average AI-generated content rate of 13.8% across platforms, with variation by platform and content length.
- Longform content (over 250 words) is particularly affected, with 25.72% fully AI-generated on average, except on Substack where longer posts are less likely to be AI.
- LinkedIn has the highest AI saturation, with over 40% of longform posts flagged as fully AI-generated and accounting for 62% of all flagged AI content.
- X/Twitter has the worst combined AI rate, with almost half of articles either fully AI-generated (23.9%) or AI-assisted (22.9%).
- Reddit has low overall AI share (4.4%) due to human-authored replies (98.1%), but top-level posts have higher AI rates (11.6%), similar to X/Twitter.
- Anti-botting strategies on platforms like Reddit may miss AI-generated top-level posts, which have more audience impact despite lower volume.
- The dataset includes over 1 million posts scanned with Pangram's AI detection model, highlighting AI writing as a widespread issue on social media.
- Transparency through tools like Pangram's extension aims to help users regain control over their attention in an internet increasingly flooded with AI content.
The Shape of Enshittification
3 months ago
- A study by University of Maryland and Google DeepMind analyzed 61,608 stories and found AI-generated writing has a distinct 'shape,' identifiable with 93% accuracy. Key differences include AI over-explaining themes, being more linear, relying heavily on bodily metaphors for emotion, referencing fewer specific texts/brands/places, and having less narrative diversity.
- The modern world excessively rewards left-brain analytical work (e.g., spreadsheets, KPIs) while squeezing out right-brain activities (meaning, beauty, awe), leading to a sense of emptiness and unreality as we try to quantify inherently qualitative human experiences.
- Cory Doctorow's term 'Enshittification' describes the predictable decline of digital platforms: starting with early adoption, scaling, critical mass, then degrading with ads and business model dominance, and finally decline as users leave.
- Social media is in advanced stages of Enshittification. Platforms like LinkedIn are filling feeds with ads and AI content, reducing organic reach. The rise of AI-generated interactions diminishes the human connection that motivates posting, pushing real sharing to private chats.
- How-to nonfiction book sales are plummeting (e.g., Tim Ferriss's sales down ~80% since 2022), as AI chatbots provide instant, tailored advice. However, books offering unique perspectives, big ideas, or personal transformation are thriving, indicating a shift from information to transformation.
- The internet is being mutated by content engineered for AI scraping rather than human readers, exemplified by companies like Shopify publishing self-serving listicles to manipulate chatbot answers, creating a feedback loop of low-quality, self-referential information.
- AI-generated 'slop' is infiltrating private conversations, with people using chatbots to craft messages, leading to impersonal and 'gross' interactions. This raises concerns about authenticity and the erosion of genuine human communication.
- The future of work is bifurcating into three areas: orchestrating AI systems, skilled trades/craftsmanship in the physical world, and human-to-human work requiring trust and discernment. Real, human-centric value will become rare and highly valued amidst an ocean of AI slop.
We Are Living in a 'ChatGPT Flyer Pandemic'
3 months ago
- ChatGPT-generated flyers are increasingly common both online and offline, leading to a so-called 'ChatGPT flyer pandemic'.
- These flyers are characterized by generic design elements such as bold text on dark backgrounds, AI-generated images, and lists of icons, often with arrows and checkmarks.
- Graphic designers, musicians, bars, and small business owners are actively criticizing these AI-generated flyers for being generic, lazy, and damaging to brand identity.
- Anti-ChatGPT flyer movements are growing globally, with stickers, bans in venues, and viral posts discouraging their use across different languages.
- Critics argue that ChatGPT flyers lack the charm and authenticity of traditional design, even compared to low-quality handmade or MS Paint-style posters.
- The rise of AI flyers is tied to the decline of mid-level design jobs, with generative AI allowing easy but soulless 'polish' without creative constraints.
AI children's books, body horror edition
3 months ago
- AI-generated children's books lack uniqueness and exhibit sameness despite improvements in models.
- The author purchased an Amazon bestseller AI children's encyclopedia, noting targeting reasons: high sales, gift purchases by non-readers, and no intellectual property risks.
- The book contained disturbing body horror imagery, including mutated creatures and unsettling scenes, raising concerns about its impact on children.
- AI-generated content is widespread on platforms like Amazon, with potential real sales despite fake reviews, highlighting regulatory needs.

ai-generated content

AI broke writing as proof-of-thought

LinkedIn and X Are Flooded with AI Spam, Browsing Data Suggests

AI Content Is Everywhere on Social Media, Especially LinkedIn

The Shape of Enshittification

We Are Living in a 'ChatGPT Flyer Pandemic'

AI children's books, body horror edition