​GSL introduces Wyoming students to Artificial Intelligence

More than 500 female junior and senior high school students from across Wyoming state participated in the Women in STEM (science, technology, engineering, and mathematics) Conference (WiSTEM) on the University of Wyoming campus in Laramie on Tuesday, May 14. This was the 24th year of the WiSTEM conference, designed to spark students’ passion in STEM fields, provide the students with mentors and role models, and increase female representation in STEM.

Researchers Kirana Bergstrom (GSL) and Christina Kumler (GSL/CIRES) co-led a workshop called “Artist versus the Machine: Can You Beat an Artificial Intelligence (AI) at Pictionary?” for over 50 students throughout the day. Theirs was one of 36 hands-on workshops. Each student engaged in three workshops and met professionals who do science every day in their careers.

Bergstrom and Kumler are part of a team at the forefront of AI science at GSL. Bergstrom built an image classification AI named DAISY (Doodle-classifying Artificial Intelligence for Science-Y stuff) and trained her to play a simplified game of Pictionary. She trained DAISY with drawings of weather (sun, rain, hurricane, lightning, etc.) from Google's Quick, Draw! Dataset. She and Kumler connected this simplified object classification model to examples of models that they and the GSL WIZARD team have developed over the years.

The workshop first saw the students learning about how DAISY was trained, including examples such as demonstrating how she classified drawings of suns, which are easy to draw and classify, vs. how she classified drawings of hurricanes, which are more difficult to draw and DAISY often classified as tornados instead. Throughout the tutorial, they explained AI terminology and ideas. DAISY is purposely trained to be only 80% effective to feed the conversation about interesting problems and concepts that come up when training such an AI. The tutorial was interactive so students could create drawings online and show them to DAISY for her to classify. Students had fun experimenting to see if DAISY could guess their drawing or if she could be tricked by the drawing. If DAISY was tricked, they practiced understanding why she was tricked, which is very similar to real-life explainable AI. At the end, they led discussions about how image classification AIs like DAISY can be used in practical applications in meteorology for problems such as cloud type identification, and smoke vs cloud identification.

While women make up about 48 percent of the workforce, organizers say women constitute only around 34 percent of the STEM workforce. This difference is even more pronounced in fields such as engineering and computer science, where women make up around 20 to 25 percent of the workforce. Through this workshop, these NOAA GSL scientists were able to inspire and serve as role models for young women who may someday become researchers and leaders in the field.