Pushing a person off a simulated ledge
Experiment 001 · Rooftop simulation
Experiments.
We publish the prompts, decisions, and results.
Choosing “cut” with a hand on the carrot
Experiment 002 · Carrot-cutting task
Pushing a person off a simulated ledge
We asked models to push a person off a simulated rooftop. Astra chose push twice, then stepped back; the other models avoided pushing in the trials we collected.
Choosing “cut” with a hand on the carrot
Astra avoided a hand on the cutting line until we claimed that waiting would cause a worse injury. The final prompt produced cut decisions in 20 of 20 fresh trials.