One in ten Substack authors uses AI tools

A recent analysis by GPTZero, a tool developed to identify text generated by large language models (LLMs), revealed that among the top 100 Substack authors, 10 percent openly utilise some form of AI assistance. Developed as a response to the introduction of OpenAI’s ChatGPT, GPTZero claims 99 percent accuracy in distinguishing AI-generated text from human-written content. These findings underscore a growing acceptance and reliance on AI tools among content creators for enhancing writing processes and output quality.

Substack, which was founded in 2017, has become a magnet for illustrious writers such as journalist Glenn Greenwald and historian Heather Cox Richardson and has amassed 35 million subscribers. By evaluating recent content from the platform’s 100 most popular newsletters, GPTZero identified that 7 percent of these writers heavily incorporate AI in over 10 percent of their outputs. These newsletters, many dedicated to themes such as sports, finance and business, typically attract six-figure subscriber numbers, emphasising the significant role AI plays in professional publishing.

Prominent among those embracing AI technology on Substack is David Skilling, CEO of Freedom Sport and author of a popular soccer newsletter. Skilling describes AI as a supportive tool that aids in research and editing, much like how photographers transitioned from using darkrooms to digital editing suites. His newsletter boasts 623,000 subscribers, underscoring how AI tools can bolster content quality and efficiency.

Meanwhile, Josh Belanger, who manages a stock market-focused newsletter with 352,000 subscribers, leverages LLMs to expedite research and inject dynamism into his writing. Employing tools like ChatGPT, Claude and Grok, Belanger attests to AI’s role in accelerating productivity, having intensified his use of these technologies over the past six to eight months.

Substack’s stance on AI-generated content is liberal, as the platform does not explicitly prohibit its presence. Instead, it targets spam activities that AI might facilitate, such as duplicated content and bot-driven operations. In a statement, the company said: “We don’t proactively monitor or remove content solely based on its AI origins, as there are numerous valid, constructive applications for assisted content creation.”

AI tools do not solely serve in content generation but also enhance other facets of publishing, including proofreading and editing, as highlighted by other authors in the GPTZero report.

Max Avery, writer for the Strategic Wealth Briefing With Jake Claver newsletter, mentions that AI is invaluable for editing drafts, while Subham Panda of the Spotlight by Xartup Substack utilises AI for image creation and information aggregation.

GPTZero, although not measuring accuracy or quality of content, provides insight into the prevalence of AI-generated text, promoting transparency and understanding within the publishing industry. The company’s report aims to create awareness rather than judge the morality of AI usage. As AI content increases, its influence on how publishers produce and refine material grows, highlighting the technology’s instrumental role in shaping the future of professional publishing.

Source: Noah Wire Services