A research group at Stanford University has issued a report indicating that large language models (LLMs) are now extensively utilized across a range of formal written communication contexts. The study delved into English-language texts sourced from prominent platforms and utilized a statistical model known as the 'Distributed LLM Quantification Framework' to gauge the prevalence of AI-generated content within a given period. It's important to note, however, that this approach is currently tailored for English-language datasets only.

The findings disclosed that, among various text categories analyzed last year, roughly 24% of corporate press releases, 18% of consumer grievances, 14% of United Nations press statements, and close to 10% of job advertisements exhibited characteristics indicative of AI generation. The researchers highlighted that LLMs have emerged as a vital support mechanism for everyday written communication, seamlessly blending into the formal text composition processes across diverse sectors.

Interestingly, the study observed that areas with comparatively lower educational attainment levels showed a higher incidence of AI-generated content in complaint submissions. In the realm of job postings from small and medium-sized enterprises (SMEs) or startups, the percentage of AI-influenced content ranged from about 10% to 15%.

Furthermore, the researchers underscored that, as generative AI technology continues to evolve, the task of precisely discerning the extent of AI's contribution to text creation will pose an escalating challenge in the years ahead.