Since the release of ChatGPT at the end of 2022, discussions on how artificial intelligence language models will affect society are heating up. The latest research shows that AI writing tools are growing rapidly in multiple fields, especially in low-education areas in the United States. A research team at Stanford University analyzed more than 300 million texts, and the results showed that AI language models currently account for one-quarter of the application proportion in professional communications.
The study looked at the adoption of AI writing tools between January 2022 and September 2024, and the data included 687,241 consumer complaints, 537,413 company press releases, 3.043 million jobs, and 15,919 UN press releases. The study found that about 18% of consumer financial complaints reflect the auxiliary role of AI, compared with 24% in company press releases. In recruitment information and UN press releases, AI application rates were 15% and 14% respectively.
Interestingly, despite the higher overall adoption rate in urban areas (18.2% vs. 10.9%), AI writing tools are used more frequently in areas with lower education levels. In areas with lower education levels than the state median, AI adoption rate reached 19.9%, compared with 17.4% in areas with higher education levels. This phenomenon is contrary to previous trends in technology adoption. Generally, people with higher education will accept new technologies faster.
In addition, the study also showed that among consumer complaints in different states, Arkansas had the highest AI adoption rate, reaching 29.2%; while West Virginia had the lowest adoption rate, at only 2.6%. The adoption of AI writing tools in different organizations has also shown a similar growth trend, with usage in all industries rising rapidly within the months since the launch of ChatGPT and flattening by the end of 2023.
The research team noted that although their analysis focused primarily on English content, their findings suggest that the actual usage of AI writing tools may be higher. The results of this study show that with the widespread use of AI writing tools, society may face more and more challenges in communication, such as doubts about the authenticity of information.