AI writing tools are shaking up the professional world, and a recent study gives us a fascinating look into this transformation. Conducted by researchers from Stanford, the University of Washington, and Emory University, this study dives into over 1.5 million texts from early 2022 to late 2024.
What they found is pretty eye-opening: AI-generated writing is becoming a staple in many fields, from handling consumer complaints to crafting international communications.
Let’s talk about press releases, for instance. AI’s mark here is undeniable. The study looked at more than 537,000 press releases from platforms like Newswire, PRNewswire, and PRWeb, and found that about 24% of the content was either created or significantly altered by AI. This trend really picked up after the launch of ChatGPT in November 2022. By December 2023, Newswire had hit a peak in AI usage at 24.3%, which then settled around 23.8%. PRNewswire and PRWeb weren’t far behind, with their peak rates hovering around 16.4%.
Technology and business sectors are leading the charge in adopting AI, but it’s not just about press releases. In consumer complaints filed with the U.S. Consumer Financial Protection Bureau, around 18% were AI-generated by late 2024.
Newer companies, especially those started after 2015, are embracing AI for job ads, with adoption rates between 10% and 15%. Meanwhile, older companies, particularly those founded before 1980, are more cautious, with less than 5% using AI.
The United Nations is also getting in on the action. Their press releases show notable AI usage, especially in Latin America and the Caribbean, where it reaches about 20%. While these advancements make writing more accessible, the researchers do caution about potential downsides, like losing some of the authenticity and uniqueness in writing.
Despite the study’s findings, the researchers suspect AI’s actual impact is even greater. Why? Because detecting AI in heavily human-edited content and sophisticated AI-generated texts that mimic human writing is quite tricky. The study didn’t explore AI’s role in areas like social media content, but an earlier study by the same team highlighted AI’s growing presence in academic and scientific literature.
They used a unique statistical approach, maximum likelihood estimation, to accurately spot AI content by comparing it with known references. This method provides a solid framework, analyzing statistical patterns instead of individual texts.
