Signaling the Adoption of Responsible AI Principles: A Study of AI Job Advertisements

Little is currently known about the extent to which Responsible Artificial Intelligence (AI) principles are implemented in organizations and the organizational processes through which these principles unfold. This theoretically informed and empirically grounded paper sought to bridge this crucial knowledge gap. It draws upon unique job advertisement data for AI-related positions from the USA, the UK, and India to investigate whether and how organizations signal their commitment to Responsible AI (RAI) principles during the recruitment process. The Natural Language Processing analysis indicates a limited presence of RAI signals across and within circa 9.000 job advertisements analyzed. Where these were present, the analysis suggests variations of RAI signals, especially across different employment statuses, and seniority levels. The findings hint that organizations predominantly consider RAI principles relevant to full-time positions for entry and mid-senior-level AI jobs. The consistency of signals varied in terms of the actual RAI keywords and their combinations. The correlation analysis provides insights into the interconnectedness of some of the RAI keywords, emphasizing the multifaceted nature of addressing ethical, legal, and quality considerations in AI development and deployment within the RAI context. We also noted the disparity between the RAI keywords used in job advertisements, and those the most frequently used in existing RAI and/or Ethical guidelines. Finally, the findings demonstrate that translation of RAI principles into practices might be challenging even for the companies recognized as the most ethical. The study has important implications for theory and practice, as well as builds an agenda for future research.