AI Applied: Covering AI News, Interviews and Tools - ChatGPT, Midjourney, Gemini, OpenAI, Anthropic Spotting AI Writing: Insights from Wikipedia's Guide
35 snips
Nov 29, 2025 Hosts dive into Wikipedia's guide on detecting AI writing, offering practical tips on identifying AI-generated text. They highlight the limitations of automated detectors and discuss common traits of AI writing, like exaggerating facts. Insightful examples from Wikipedia illustrate these points, and they critique the blandness of AI-generated press releases. Conor shares his sensible approach to integrating AI in his workflow, while practical advice on refining AI text ensures a more human touch. Tune in for a thoughtful exploration of AI's influence on writing!
AI Snips
Chapters
Transcript
Episode notes
Wikipedia's Field Guide For AI Writing
- Wikipedia created a battle-tested field guide for spotting AI writing used by editors to triage millions of edits.
- The guide reveals repeatable patterns in LLM output that humans can learn to recognize quickly.
Comparing Models With AIbox
- Conor uses multiple models and pays for different services to compare natural-sounding outputs daily.
- He finds model quality varies and praises tools like AIbox.ai for side-by-side testing.
AI Hypes Mundane Details
- LLMs tend to overemphasize importance and add hype to mundane subjects using repeating clauses.
- This “undue emphasis” pattern shows up across many AI-generated articles and edits.