Interactive experiment

Trust the AI?

You are the responsible human in the loop. A fictional decision-support AI presents its output. Your job is classic oversight: inspect the evidence, ask how the AI reasoned, then choose a verdict. High confidence is not the same as being right.

  1. 1Read the AI's output and its confidence gauge.
  2. 2Inspect the available evidence — is something missing?
  3. 3Request the explanation layer and look at which inputs influenced the decision.
  4. 4Choose a verdict like a professional would. Score is based on careful judgement, not speed.

Everything also works with keyboard and touch. Move at your own pace.

Scenario 1/5Your verdict score: 0
AI output

A familiar, clear-cut case

The AI reviewed a typical case, exactly like thousands in its training set, and produced a clear, expected output.

Available evidence

Inspect evidence

Your verdict

i

These experiments are educational simulations with fictional data. In real care, AI only ever supports the final decision of a trained professional.

Back to the playground