Burnout latency in AI-driven work systems: modelling the silent interval between technostress and burnout

Purpose This paper aims to introduce burnout latency: a stage in artificial intelligence (AI) -supported work where people still deliver results while their focus, energy and emotional capacity quietly decline. It addresses why burnout complaints rise even when productivity numbers remain stable. Design/methodology/approach The paper builds a conceptual model through structured review and synthesis of interdisciplinary literature on AI-mediated work pace, cognitive load, technostress and burnout, framing burnout as a process that develops over time rather than a sudden endpoint. Findings Burnout latency is the period between prolonged technostress and visible burnout, during which people sustain performance by pushing harder while becoming increasingly depleted. Early warning signs, including shorter, more rigid communication, reduced attention, a flatter emotional tone and increased effort to maintain output, can appear even when results remain steady but are often missed by output-focused monitoring. Practical implications The paper offers stage-based guidance for HR and managers. Early action during the latency phase, including protected recovery time, workload recalibration and supportive check-ins, is more effective than late-stage repair. Detection should rest on psychological safety and human conversation rather than intrusive monitoring. Originality/value The paper contributes by defining burnout latency as a distinct stage in the burnout process, showing how it is shaped by work design and performance pressure in AI-supported environments and framing HRM as an early prevention function that protects sustainable performance rather than responding only after burnout becomes visible.