The proliferation of AI large language models (LLMs) is transforming communication across business, consumer, and institutional domains, as highlighted by research led by Weixin Liang at Stanford University.

The usage of AI-generated content surged significantly after the launch of ChatGPT in late 2022, particularly among smaller firms, with AI-assisted job postings reaching 15%.

By the end of 2024, studies showed that AI-assisted writing was present in 24% of business press releases, 18% of financial grievances, 10% of job postings, and 14% of UN press releases.

International organizations, including the UN, have incorporated LLM usage in their communications, reaching about 14% in press releases.

A significant issue is the potential for over-reliance on AI-generated content leading to generic writing that lacks unique voices, especially in job listings.

Liang's research indicates that if many communications are AI-generated, it could foster public skepticism regarding their authenticity and origin.

While AI writing enhances efficiency and aids non-native speakers in expressing themselves, it raises concerns about authenticity and communication quality.

The study also warns of recursive feedback loops, where future LLMs may be trained predominantly on AI-generated content, risking the loss of accuracy and creativity.

'Model collapse' is a phenomenon where AI models trained on their own outputs degrade over time, producing unreliable and repetitive content.

Future research initiatives are planned to assess the impact of LLMs on communication in high-stakes contexts, acknowledging that AI writing represents a fundamental shift in mass communication.

Liang's research, one of the largest empirical investigations of AI writing adoption, analyzed over 300 million online documents and posts from 2022 to 2024.

As AI-generated content becomes more common, businesses will face challenges in balancing efficiency with authenticity and credibility in communication.