8th grade students explore artificial intelligence

As artificial Intelligence becomes a routine presence in our lives, 8th grade English students have begun investigating how AI creates and sorts natural human language. Students began by looking at artificial intelligence broadly, and took a homemade Turing Test to see whether they could tell the difference between human and AI artists, musicians, and writers. Then we took a deep dive into StoryQ, an AI text classification model that took us behind the scenes to see how a machine "reads" human text, identifies key words, and assigns them negative or positive weights to make predictions about the overall sentiment, or tone.

Students fed StoryQ ice cream reviews, then analyzed the words the machine used to make its predictions about sentiment. Some of the machine's decisions surprised us. For instance, it assigned a heavy negative weight to "like" and "just," which were sometimes used positively. It was also flummoxed by sarcasm :). At first, StoryQ was only able to predict 78% of reviews accurately, so we analyzed error cases and proposed new language rules to improve the model.

Once we understood more about how AI generates and classifies text, we turned our attention to its limitations and its role in our lives. We looked at news stories and discussed whether AI is sentient, and whether it will ultimately improve or worsen human communication, and human life. Students raised concerns from disinformation to job loss. Below is an AI-generated image of 8th graders having a fishbowl-style discussion, courtesy of DALLE-2, and an actual image of our class doing the same thing.