Bogdan Vrusias

“Human in the loop” emerged as a human centric safeguard: a way to keep people visibly involved in AI workflows and preserve human judgement at critical points. But as AI systems become increasingly agentic and capable of shaping the workflow itself, this model often reduces the human role to validation, monitoring, or procedural compliance. In this session, Bogdan Vrusias, Global Head of AI and Data Engineering at The Economist, examines how HITL can unintentionally narrow human agency rather than reinforce it. He argues for reframing the loop so that AI sits within a broader human process, one grounded in contextual reasoning, editorial discretion, and genuine ownership. The session offers a more honest and future proof approach to ensuring that humans guide the system, not the reverse.

• Revealing how HITL often constrains, rather than expands, human decision space

• Reframing AI as embedded in human processes, not humans inside machine loops

• Designing agentic-era workflows that preserve judgement, context, and real agency