The Slop Index: Which AI Model Writes the Least Like AI? (www.theslopindex.com)

A new tool called the Slop Index has been introduced to evaluate how closely AI-generated text resembles genuine human writing. By assessing 39 frontier AI models against a curated dataset of authentic human communications from the early 2000s, this live benchmark aims to categorize models based on their "sloppiness," with higher scores indicating writing that feels more robotic. Users can participate in blind voting to help determine which models exhibit less human-like qualities, focusing on linguistic elements that differentiate chatbots from actual human correspondences. This development is significant for the AI/ML community as it addresses a critical challenge: distinguishing between human-written and AI-generated text. By employing a methodologically rigorous approach—using pre-ChatGPT data as a baseline—the Slop Index offers a quantifiable metric for evaluating AI performance. This may influence model training and usage, encouraging developers to create more human-like interactions in AI applications and establishing a clearer understanding of the current landscape of generative AI writing capabilities.