Human writer or AI? A detection model being developed at Stanford

 

The Human Centered Artificial Intelligence program at Stanford University reports that scholars there have built a detection program that "...can determine with up to 95% accuracy whether a large language model wrote that essay or social media post."  For more information on their efforts, use this link to the news release.

After a somewhat long hiatus in posting to my blog, I'm planning to follow developments relating to the public release of chatGPT and competing programs soon to be released by Microsoft, Facebook (Meta) and others.  This Spring semester I'm teaching both an undergraduate course on digital networks and innovation and a graduate seminar on networked communication technologies, so the development and spread of generative natural language programs is highly relevant in both classes.