Agent evaluation, simulation environments, long-horizon decision-making, and autonomous businesses
Can AI learn from delayed consequences?
