Reinforcement Learning Agent Eval
A set of 10 RL questions spanning topics like RL agent training in Minecraft, Rocket League, etc.
Prompt
Analyze the emergent behaviors and cognitive limitations of a Reinforcement Learning (RL) agent within a standard Minecraft survival environment across the following parameter milestones: 1, 10, 100, 1k, 10k, 100k, 1M, 10M, 50M, and 100M. For each milestone, please provide: Behavioral Profile: Describe the agent's primary interaction with the world (e.g., motor noise vs. goal-oriented pathfinding). Cognitive Ceiling: Identify the most complex task the agent can realistically achieve (e.g., "Breaking a block" vs. "Navigating the Nether"). Human Perspective: Map the agent’s skill level to a human equivalent (e.g., 'Static Object,' 'Toddler,' 'Casual', 'Speedrunner').
