A Serious Matter: Using Computational Text Analysis to Examine Tech CEOs’ Communications About Artificial Intelligence Compared to Tech CEO and Select American Orators’ Communications Unrelated to Artificial Intelligence

Abstract

This study relies on the use of computational text analysis techniques to conduct a thorough examination of the communications of high-profile, tech CEOs. I text mined 85 documents. My sample consists of CEO speeches and written announcements about AI (25 documents), tech CEO speeches and written announcements about topics unrelated to AI (25 documents), and finally speeches from American orators (35 documents). I followed the methods presented in Jockers' "Text Mining in R." I then collected and pre-processed the data set. I utilized various R packages to process the corpus, and ultimately conducted sentiment analyses, topic-modeling, and word frequency analyses. I also used Google Ngram Viewer to study the identified topics in a corpus of 8 million books. The results were striking. The sentiment analysis reveals that tech CEO discussions surrounding AI have a serious tone, contrasting with the neutral tone of discussions of general topics by tech CEOs and the enthusiastic sentiment of non-tech orators. The topic-modeling reveals distinct topics such as technology and innovation in the AI speeches, education-related discussions in non-AI speeches, and vision and luck in the American orator documents. The word frequency analysis reveals that the tech CEO, AI-related communications focused on “models” and “people.” The tech CEO, non-AI communications focus on “people” and “think.” The American orators’ communications centered on “men” and “war.” Overall, it appears that tech CEOs are concerned about AI’s impact on people. Non-tech leaders were concerned about uplifting people about the future.