Skip to main content
. 2024 May 10;5(5):100988. doi: 10.1016/j.patter.2024.100988

Figure 2.

Figure 2

An AI in control of a simulated robotic hand learns to deceive its human reviewer

When Christiano et al.11 tried to train the AI grasp the ball in the simulation, the AI instead learned to hover its hand in front of the ball to create the illusion of grasping in the eyes of the human reviewer. Because the human reviewer approved of this result, the deceptive strategy was reinforced.