Embodied AI fails basic spatial reasoning
A new testing method found that leading AI agents make massive spatial reasoning errors, scoring half as well as humans.
A new testing method found that leading AI agents make massive spatial reasoning errors, scoring half as well as humans.