[ESSEC Knowledge] by Valery Yakubovich - Associate Professor of Management at ESSEC Business School and Senior Fellow at the Wharton Center for Human Resources.

Abstract

The rapid evolution of digital transformation in management, particularly in the context of AI, poses substantial challenges in implementing artificial intelligence techniques in human resources management. This article, authored by ESSEC management professor Valery Yakubovich and his colleagues Peter Cappelli and Prasanna Tambe from the Wharton School, identifies four primary challenges in utilizing AI in HR and offers practical responses.

The challenges include:

1. Complexity of HR problems: HR issues, which involve human interactions and behaviors, are nuanced and multifaceted. Data is not always readily available, and ethical concerns emerge, requiring careful consideration.
2. Small datasets: HR datasets are often smaller than those in other domains, and data science techniques may not perform well in predicting rare outcomes. Ensuring sufficient data for AI algorithms is essential, and thoughtful data aggregation and integration are recommended.
3. Ethical and legal considerations: AI in HR may lead to ethical dilemmas, particularly in areas like hiring. Legal norms related to disparate treatment should be taken into account when developing AI algorithms.
4. Employee reactions: Shifting decision-making processes from humans to algorithms can have repercussions on employee attitudes. It's important to consider how employees will react to AI-driven decisions.

Throughout the AI project life cycle, from data generation to decision-making, these challenges persist. Causal explanations and ethical considerations are vital in HR analytics and decision-making. Embracing HR algorithms' limited predictive power and using randomization for decision-making can be effective strategies. Formalizing the algorithm development process and involving stakeholders can help ensure that employees understand and accept AI-driven outcomes.

[To read the full article please follow this link.]