A Wikipedia Group Made a Guide to Detect AI Writing. Now a Plug-In Uses It to ‘Humanize’ Chatbots (www.wired.com)

The introduction of a new plug-in called "Humanizer" has garnered attention within the AI/ML community for its innovative approach to refining chatbot interactions. Created by developer Chen and hosted on GitHub, Humanizer utilizes a guide from WikiProject AI Cleanup, which identifies 24 linguistic patterns typically associated with AI-generated text. By informing Anthropic's Claude coding assistant to avoid these "giveaways," the plug-in aims to make outputs feel more human-like. Notably, Humanizer encourages Claude to express opinions and react emotionally, distinguishing its responses from the precise and sometimes clinical style common in AI outputs. While the initiative is viewed as a creative solution to lowering the detectability of AI writing, it does come with potential drawbacks; the adjustments might sacrifice factual accuracy and coding proficiency. As AI writing detection remains a challenge—where even human-written texts may carry detectable "AI traits"—this tool reflects an ongoing struggle within the community to balance human-like expression with substance and reliability. The existence of this plug-in underscores the complexities of AI writing, and its development might inspire further exploration into nuanced detection methods that go beyond surface-level linguistic patterns.