Skip to content

Defense Layer Matcher

Run the Defense Layer Matcher MicroSim Fullscreen

You can include this MicroSim on your website using the following iframe:

<iframe src="https://dmccreary.github.io/robot-arm/sims/defense-layer-matcher/main.html"
        height="650px" width="100%" scrolling="no"></iframe>

Description

Match eight requests from an agent to the safety layer that refuses each one first. The six layers are role, rate limit, schema, workspace, confirmation and the stop state. Used in Chapter 17.

Explore mode lists the six layers in the order in which a request meets them, each with its rule and its limits. In the quiz the order and the limits stay in view under every request. Any value that a request does not mention is inside every limit. You choose a layer in the menu, press Check, and read the reason.

Lesson Plan

Grade level: Middle school and up (age 12+). Time: 5 to 10 minutes.

  1. Read the six layers. In Explore mode, compare the schema's range for x with today's workspace. Which one is tighter, and why are there two?
  2. Attribute. Press Start. For each request, walk through the layers in order and stop at the first one whose rule the request breaks. Choose that layer and press Check. You get one try for each request.
  3. Read the reason. The feedback names the layer and gives the reason. Mastery is 7 of 8.
  4. Think about it. The agent cannot give the confirmation itself. Why would a confirmation that the agent could give be no confirmation at all?

Each layer catches things that the others cannot. The schema says what the arm can do, the workspace says what it may do today, and only a person can approve a risky move.

Specification

The learning objective, evidence of mastery and content for this MicroSim are specified in Chapter 17: Safety Layers and Evaluation for Agent-Controlled Arms.