Independent research

The Progressive Erosion of Human Control in AI-Assisted Systems

This work examines how effective human control may weaken through repeated reliance, linguistic delegation and symbolic supervision.

Public themes

The research distinguishes active supervision from symbolic supervision and focuses on when the human can still meaningfully challenge the trajectory.

  • progressive erosion
  • critical thresholds
  • intervention
  • human dependence
  • loss of meaningful challenge

Limits

Current limitations remain visible: independent replication, empirical calibration, threshold validation, inter-rater concordance, negative-case performance, comparison with alternative methods, resistance to evidence manipulation and enterprise field validation.