The competition focused on using advanced machine learning techniques to predict future candidate performance while eliminating bias in hiring decisions
Axiom Consulting Partners (“Axiom”), an advisory firm integrating strategy, artificial intelligence (AI) and behavioral science to help clients grow and transform their businesses, has been announced as a winner in the third annual Machine Learning Competition hosted by the Society for Industrial and Organizational Psychology (SIOP). The 2020 and 2021 SIOP competitions were combined due to the COVID-19 pandemic and received entries from more than 200 teams—comprised of 500 professionals from industry and academia—across the entire data science community, including psychologists, computer scientists and engineers from various disciplines.
This year’s competition focused on how AI and machine learning could be leveraged to solve one of the most common challenges HR departments and hiring managers face today: the ability to make fair and accurate hiring decisions. SIOP competition organizers challenged participating teams to design algorithms that maximized the accuracy of predicting candidate performance and retention while minimizing bias.
Axiom’s team of psychologists, economists and computer scientists created numerous methods for scoring employee data to engineer a robust modeling dataset and arrive at an optimal solution. The Axiom team will present their approach and solutions alongside three other winning participants at a special session during the 36th Annual SIOP Virtual Conference on Thursday, April 15th.
Scoring was based on how well participating teams demonstrated ability to balance accuracy in predicting job success and retention while minimizing bias against protected groups. Each team was tasked with the development of a unique algorithm that leveraged applicant pre-hire assessment data and post-hire performance and turnover data from 50,000 employees of a Fortune 100 company.
“I’m really proud of the work our team accomplished in developing fair AI that can be scaled to help solve real-world challenges that society and our clients are grappling with every day,” said Dr. Aaron Sorensen, a Partner at Axiom who leads the firm’s Business Transformation and Behavioral Science teams. “We have the ability to bring cutting edge data science and behavioral science capabilities from both industry and academia together to solve business problems, which is not only extremely exciting but a sign of the high-impact, high-reward innovation that can happen when analytics is applied to the most foundational parts of an organization.”
“Our team experimented with various approaches to the modeling system leveraging machine learning and combinatorics. Key contributors to the success of our approach were techniques used for the imputation of missing data and the synthesis of new features,” said Ian Burke, a Senior Data Scientist with Axiom. “The algorithm recommended hiring decisions based on the probability of individuals being high performers. To achieve a fair hiring sample, the team substituted out the lowest performers in favor of those who were more likely to be in the protected group.”
Added Sorensen, “Selecting the right person for the job is a challenge that goes back to the beginning to the field of Industrial Psychology, when the military needed to quickly select officers in wartime. While significant strides have been made in predicting future success on the job, the well-documented disadvantages inflicted on minority groups and disenfranchised communities along the way cannot be ignored. What this competition shows is that there is potential for AI tools to provide strategic guidance and data-backed solutions to institutions and businesses looking to make the best hiring decisions while minimizing bias. That’s a big deal!”
