AI’s Impact on Professional Writing
A recent study from Stanford University researchers shows that large language models (LLMs) have rapidly affected how we write across multiple domains of society—often in ways we may not fully appreciate.
This systematic analysis of LLM-assisted writing examined millions of samples across four distinct domains, pointing to a fundamental shift in professional communication practices with implications for AI literacy education.
Among the findings...
Widespread adoption across diverse domains:
By late 2024, LLM-assisted writing had penetrated multiple sectors of society at significant levels - approximately 18% of financial consumer complaints, up to 24% of corporate press releases, nearly 14% of UN press releases, and up to 15% in job postings - mirroring the widespread adoption seen across academic researchers shown in previous research.
Consistent adoption pattern:
All domains showed a similar trajectory - minimal LLM usage before ChatGPT's release in November 2022, followed by a rapid surge 3-4 months after its introduction, then stabilization by late 2023, suggesting either market saturation or perhaps more likely the inability to accurately capture increasingly sophisticated LLM usage.
Organizational characteristics influence adoption:
Smaller, younger organizations (particularly those founded after 2015) demonstrated significantly higher adoption rates, with differences also observed across sectors (Science & Technology showing the highest rates) and geographic areas (higher adoption in urban environments).
As writing increasingly becomes a hybrid human-AI activity, how will our educational systems need to evolve?
These domains primarily involve straightforward writing tasks, but as both development and adoption continue to advance, understanding the effects on quality, reliability, and creative expression while simultaneously navigating the regulatory and ethical landscape becomes critical.
If you're interested in reading the complete study, the paper can be found here.
