Navigating the New Age of Writing: Tools to Detect AI Influence

As artificial intelligence continues to permeate various writing forms, from LinkedIn updates to academic papers, writers are increasingly concerned that their own work may inadvertently reflect an AI-like style. This phenomenon has prompted a wave of innovative tools designed to help authors identify and rectify any AI-infused language in their writing.

Both LinkedIn and Substack have recently taken steps to minimize the prevalence of fully AI-generated content, while AI detection platforms like Pangram and Turnitin are being utilized by universities and organizations to scrutinize everything from application essays to workplace documents. However, the challenge now extends beyond detecting AI text; writers who do not engage with AI tools are finding themselves anxious about accidentally incorporating AI tendencies into their prose.

George Strakhov, founder of the AI automation consultancy move38, emphasizes the unique characteristics of AI writing. "These models do speak in a specific way, and you develop a sense of it—you develop the smell of the writing," he explains, highlighting the importance of distinguishing between human and AI-generated content.

To tackle this issue, Strakhov and other designers have created tools that analyze writing for common AI markers. Among them is the LLM cliché highlighter, developed by programmer Simon Willison. This tool, launched on August 28, allows users to paste their text and identifies 38 distinct signs of AI-generated language. It highlights problematic phrases and provides a summary of flagged sentences, helping writers see how their work might echo AI patterns.

While the LLM cliché highlighter offers a straightforward analysis, its limitations lie in its static nature—it does not adapt to the evolving landscape of AI writing styles. In contrast, Strakhov's tool, Slop Tells, is designed to continuously update its criteria for what constitutes an AI tell, reflecting the dynamic nature of AI writing.

Slop Tells operates by comparing contemporary AI-generated text with human writing from pre-2022. It uses a three-pile system to assess language trends, making it capable of identifying phrases that have fallen out of favor and those that have gained traction over time. For instance, the term "delve" was once prevalent in AI outputs but has since been phased out as AI models evolve.

What sets Slop Tells apart is its ability to highlight not just AI jargon but also to indicate whether certain expressions are currently trending in AI models or have become obsolete. This tool provides users with a richer context for understanding the language they use, linking to additional resources that elaborate on each identified phrase.

As AI continues to shape our writing landscape, Strakhov cautions that exposure to AI-generated content can alter a person's taste in writing. "What you’re exposed to is what forms your taste," he remarks, urging writers to consciously navigate their interactions with AI tools to preserve their unique voice.

Source: www.fastcompany.com