idle

The world

What's actually there. Visited cells revealed unless teacher view is on.

Agent's knowledge map

What the agent has inferred from percepts so far.

Inferred safe
Inferred deadly
Possibly deadly
Unknown
Score
0
Step
0
Carrying
-
Arrows
1
Status
alive
Score breakdown: (no actions yet)

Reasoning log

🔎 Why this cell?: click the knowledge map

Hover over a cell in the knowledge map (right). The agent's inference chain for that cell appears here.

Click any cell on the knowledge map to see the inference chain.
🎯 Try these scenarios first

Pick a scenario, predict what should happen, then apply it and compare the result with your prediction.

Scenario 1: Pure logic moves only into proven-safe squares

Core question: What can a cautious agent do when the evidence does not prove a safe next move?

Run and watch: Use the pure-logic strategy on a fresh hidden world and step until the agent stops or exits.

Predict first: Predict whether stopping means every remaining square is dangerous.

Explain afterward: Explain the difference between lacking a proof of safety and proving danger.

Scenario 2: The hybrid agent can shoot, gamble, or leave

Core question: When should an agent act under uncertainty instead of waiting for certainty?

Run and watch: Switch to the hybrid strategy and compare its actions with pure logic on the same type of world.

Predict first: Predict which new action will appear when the safe frontier is empty.

Explain afterward: Explain what extra assumption or risk rule allows the hybrid agent to continue.

Scenario 3: Probability ranks unresolved risks

Core question: How can an agent choose among several squares that are all uncertain?

Run and watch: Use the probabilistic strategy and inspect the estimated pit probability for frontier cells.

Predict first: Predict whether the agent should choose the most familiar square or the lowest estimated risk.

Explain afterward: Explain how ranking uncertainty differs from proving a square safe.

Scenario 4: Reveal the world and compare truth with belief

Core question: Why can the agent’s knowledge differ from the hidden ground truth?

Run and watch: Turn on the teacher view, step slowly, and compare hidden hazards with the knowledge map.

Predict first: Predict one fact that is true but the agent cannot yet know.

Explain afterward: Use the reasoning log to identify exactly which percept or rule is missing.

Scenario 5: Unknown does not mean safe

Core question: What evidence is required before a square can be labeled safe?

Run and watch: Hide the world, use pure logic, and stop when the agent reaches an unresolved frontier.

Predict first: Predict which neighboring percept would be needed to rule out each hazard.

Explain afterward: Explain why absence of knowledge is not evidence of absence.

✨ Featured comparison
Featured comparison

Add percepts until the agent knows a square is safe, then contrast that with a square it merely suspects. Logic distinguishes certainty from risk.

Step through a board and separate guaranteed-safe moves from probability-shaped guesses.

🧭 Visual explanation

Live state to watch 
No state captured yet. Run or adjust the applet first.
Misread to avoid

📚 Lesson tour (5 steps)
Step 1/5: -

-

📝 Worksheet (3 questions)

Q1. "Pure logic" strategy. Generate several new worlds and let the agent run. Does the pure-logic agent ALWAYS retrieve the gold?

Q2. Switch to "AIMA hybrid" strategy. Compared to pure logic across many runs, the score variance (some win big, some die) is:

Q3. What is the per-step score penalty in the AIMA-style scoring (just the magnitude, no sign)?

🧪 Student response packet

Run an agent strategy, inspect the world and knowledge map, then explain how percepts support safe action.

State snapshot appears here.
♿ Text and keyboard support

Text and keyboard support

Keyboard path

  • Use Tab and Shift+Tab to move through controls.
  • Use Enter or Space on buttons, and arrow keys on sliders or select controls.
  • The visual region is focusable and described by the state summary below.

Text state summary

Reduced motion and non-visual support

Reduced-motion settings are honored where possible. The current state is also available as text, so the main result does not depend only on color, animation, or spatial position.

Accessibility note: this layer gives a text equivalent for the applet state. It does not replace a full human screen-reader audit.

↓ Read the explanation