How AI Can Evaluate Job Performance and Leadership Potential Through Emails

Assistant Professor Mustafa Akben of the Martha and Spencer Love School of Business at Elon University has developed a machine-learning model that can evaluate job performance and leadership potential simply by analyzing emails. Akben participated in the SIOP 2023 Machine Learning Competition, which challenged participants to create an AI model that could identify job candidates with leadership skills based on their email responses to fictional workplace scenarios.

The competition focused on natural language processing, a subfield of artificial intelligence that enables computers to understand, interpret, and generate human language, which has been in the headlines recently due to its use in ChatGPT. In the competition, job candidates were asked to respond to fictional scenarios they received during a daylong simulation, and their responses were scored by a panel of human judges based on how they analyzed the scenarios, the tone, and the content of their replies.

Akben’s AI model performed better than two dozen other entries and was more accurate than human judges in predicting how the second batch of candidates would be rated by the judges. This achievement was presented at the 2023 SIOP Annual Conference in Boston.

However, Akben noted that AI is not yet ready to replace human assessments of job performance and leadership traits. While his model was the most accurate at predicting how candidates would be rated on average, the resulting effect size was only moderate. To improve AI’s accuracy, Akben said that bias must be eliminated, and the effect sizes must grow higher.

Despite this, Akben is enthusiastic about incorporating emerging AI technologies into his classroom, and his students are fascinated by these technologies. He is eager to collaborate with Elon University students or faculty members on AI-related projects involving workplace or educational applications, creativity, and proactivity.