Mar 25, 2025, 12:00 AM
Mar 24, 2025, 12:00 AM

Stanford study reveals alarming rise of AI-generated writing online

Highlights
  • A Stanford University study examines the widespread use of AI in various types of written communication.
  • By 2024, the study projects that a significant portion of business reports and job listings will include AI-generated content.
  • The increasing reliance on AI challenges authenticity and raises concerns about cognitive independence.
Story

In late 2023, a significant study conducted by Weixin Liang at Stanford University highlighted the pervasive adoption of AI-assisted writing in numerous fields, including corporate communications, job listings, and international reporting. The research data indicated that by the end of 2024, as much as 24% of corporate press releases, 18% of financial grievances, and 15% of press releases from the United Nations were expected to involve AI-generated content. This marked a dramatic increase that followed the rapid proliferation of generative AI tools following the release of ChatGPT in late 2022, suggesting an accelerating trend across diverse areas of written communication. The examination of over 300 million online postings across various sectors revealed noteworthy statistics, such as 10% of job listings on LinkedIn being AI-generated and 15% of such communications originating from smaller companies. Liang's research employed specialized methods to quantify and analyze the frequency of AI-written text, showcasing the profound significance of generative AI's impact on formal communications. This rapid expansion presents new challenges regarding the authenticity and reliability of information as AI's role in content creation continues to grow. Moreover, the findings raise concerns about the dynamics of AI training. The study suggests a potential risk associated with AI models increasingly relying on content that is itself AI-generated. Such recursive feedback loops may lead to a phenomenon known as “model collapse,” where future AI language models are primarily trained on synthetic text rather than diverse human-generated material, potentially eroding the quality of AI-generated content over time. This situation poses significant implications for industries that depend on AI-assisted writing, ranging from business communications to journalism and academic research. As organizations regard AI-generated content as a standard practice, a delicate balance must be struck between efficiency and authenticity. While these technologies offer considerable productivity advantages, dependence on AI without critical oversight risks undermining creativity and credibility in essential communications. The study warns of a potential decline in intellectual independence, as reliance on AI tools may foster “agency decay” in individuals' cognitive processes. As society adapts to these changing dynamics, questions arise about the broader impact of AI on human cognition and whether it heralds an era of cognitive evolution or erosion.

Opinions

You've reached the end