- If consciousness is evolutionarily useful, do we get an easier path to AGI if we model it in a way that seems like it'll be conscious? - Build artificial models of consciousness (IIT, global workspace) to see its behavior under different circumstances