MIT Sloan Management Review and Boston Consulting Group convened an international panel of AI experts in 2025 to examine responsible AI (RAI) governance, focusing on agentic AI that exhibits humanlike traits such as behavior, communication, empathy, and persuasion. Nearly 80% of experts agree that responsible governance involves questioning the necessity of deploying overly humanlike AI systems, emphasizing not just how but when to deploy such technologies. Panelists suggest considering simpler alternatives, advocating transparency, and proceeding with caution due to risks including manipulation, emotional attachment, and overreliance. Humanlike AI may be valuable in contexts like education, therapy, and caretaking but requires deliberate design, testing, and continuous monitoring to avoid negative social impacts. The article concludes with recommendations for organizations on assessing need, educating developers, and maintaining transparency to ensure responsible adoption of anthropomorphic AI.
Back
