Technology like artificial intelligence (AI) may seem mystifying to many
AI is often thought of as a sophisticated, high-level technology. However, Jiang, whose research focuses on broadening access to STEM in areas like artificial intelligence, data science and computing, says it is really very human-oriented.
For example, in the subfield of artificial intelligence known as machine learning, the computer learns patterns from training data collected and generated by humans and then uses these patterns to make predictions based on knowledge gained from the data set.
Bringing Diversity into the Field
Biases in AI technology, Jiang said, may not be intentional but can come from a lack of diversity in the field and have negative impacts on those who use the technology.
For example, studies show that facial recognition AI technology on most smartphones works more often and more successfully for white men than it does for people from other racial backgrounds. This is a result of the training data – which contained a disproportionate number of images of white males – that was used to develop the AI model.
Exposing even those students who don’t express interest in pursuing a career related to artificial intelligence is beneficial, Jiang said, because they can learn skills related to the efficient use of technology and critical thinking about when advanced technologies should and should not be used.
“They have to have a critical perspective about how to use technology in the field they are interested in,” she said.
Jiang is helping to expose high school students to artificial intelligence through her work on the “Narrative Modeling with StoryQ: Integrating Mathematics, Language Arts, and Computing to Create Pathways to Artificial Intelligence Careers” project, which began in 2020.
The StoryQ project allows students to engage in the process of developing test classification models and question how the decisions they make along the way affect what the models look like. They must also think about what goals they have for their models and consider who would be impacted by their AI model.