Brian Mathews at Carnegie Mellon University has shared a series of really interesting blog posts where he uses a corpus of material derived from recent CNI meetings in a wide variety of ways. He tries to predict what topics will be featured at the next CNI meeting; He has some lovely visualizations from presentations. And he uses a ChatGPT based approach to analyze the materials and mine for insights.

This work is fascinating on several levels. Of course, because he's using the CNI materials as a case study it's of obvious interest to us at CNI, and perhaps to our broader community. It further reinforces our commitment to the value of capturing our conferences and making these materials broadly available. But it also suggests new ways to more efficiently and effectively make use of the overwhelming wealth of conference material that started to appear during the pandemic and has in many cases continued to be available. There are more recordings that any of us will ever have time to watch; more transcripts than we'll ever have time to skim.

Here are links to some of the recent posts

https://www.brianmathews.io/blog/my-personal-gpt-prototype-cni-insights-agent

https://www.brianmathews.io/blog/visually-representing-conference-presentations-examples-from-cni

https://www.brianmathews.io/blog/can-i-use-chatgpt-to-predict-the-next-cni-meeting

Clifford Lynch
Director, CNI