12/01/2023
Live Webinar
Speaker : Prof. Ishita Chakraborty, Wisconsin Business School
Abstract: We study the problem of AI and AI-human based screening (eliminating bottom candidates) and selection (hiring top candidates) in salesforce hiring. Using videos of structured interviews and ratings from multiple recruiting experts on standard performance criteria, we develop an AI model of salesforce “skill” prediction by extracting theory-relevant objective measures of interviewee performance from videos. Using the model, we address two issues: First, to aid interpretability, we assess what mode of unstructured data from the interview (text, audio and video) and what specific behaviors (e.g., body language, conversation style) drive AI predictions of salesperson’s ability. We find that while screening decisions are heavily impacted by audio-visual features like “energetic voice” and “open posture”; selection relies more heavily on textual content and linguistic features such as being precise and quantitative. We also find that there is an optimal amount of hand gesturing, body posture movement and conversational interactivity for success in selling. Second, we consider a hybrid AI-human model where we augment AI predictions with human interventions in a Bayesian framework. The hybrid model improves accuracy but increases cost (of human labor). Hence, we propose a cost-effective way of deploying AI in salesforce hiring with humans in the loop — use AI for screening and augment it with human judgments based on early stages of interview for selection
About the Speaker: : I am an Assistant Professor in Marketing at Wisconsin Business School. Prior to this, I got my M.A., M.Phil and PhD degrees in Quant Marketing from Yale University, School of Management. I have an MBA from Indian Institute of Management, Calcutta and a bachelor's degree in Computer Engineering from Univ of Mumbai. My research interests are in digital marketing, online platforms, text and video analytics and mobile apps. My research aims at developing algorithmic market research tools to derive richer, accurate and real-time insights from unstructured data. I use natural language processing, machine learning, deep learning and econometric modeling in my current work. Substantively I am interested in word of mouth, negotiations and brand positioning
