In the pilot study for our Reachy Mini interaction system, only 38% of participants could distinguish autonomous behaviour from human teleoperation. The knee-jerk reaction is to celebrate the 62% fooling rate.
The interesting data is in how participants guessed. Many of them assumed autonomous behaviour would feel more lifelike, and that teleoperation would feel repetitive and constrained. The exact inverse of what our system produced: our autonomous VLM tended to select generic, repetitive expressions.
The VLM recognized the action in each keyframe. What it couldn’t do was infer what the participant wanted next, or chain behaviour across turns. No persistent scene model, no anticipation.
This is the micro-version of the argument in World models as shared substrate: you can’t pick a good response if you don’t have a shared model of the thing you’re responding to.