When AI Says Yes and You Need to Say Why | Tomorrow Ready

← Back to Evaluating AI Output

Tomorrow Ready · Evaluating AI Output

When AI Says Yes and You Need to Say Why

Subject adaptation · Years 4–6 · Science · Field-Based STEM · Tony Jones

The tool most likely to validate a student’s prediction is the one they are most likely to ask. When AI consistently confirms what students already think, the productive friction that drives scientific thinking never arrives.

Phase 1
Predict Without Help
Phase 2
Name the Friction Point
Phase 3
Commit Before Investigating
Phase 4
Compare After Evidence

The Strategy

  1. Present the science task or question to the class.
  2. Students write their prediction using only what they already know. No tools, no resources, no peer consultation at this stage.
  3. Students name one reason their prediction might be wrong. This is the friction step and must be completed before any investigation begins or equipment is collected.
  4. Students commit in writing: accept the friction and describe how they will address it, or reject it with a specific reason. Both responses are valid. The commitment statement travels with the prediction throughout the task.
  5. Investigation proceeds using whatever tools the task permits, including AI where the teacher has allowed.
  6. At submission, students complete a Position Comparison: what the evidence showed, what changed, what stayed the same, and what caused any shift.

In Practice

Years 4–5

Focused on observable, concrete phenomena. “My plant will grow taller in the sunny window.” Friction step: “But maybe water matters more than light.” Commitment: “I will give all plants the same amount of water so I am only testing light.” The commitment shapes the method before investigation begins.

Years 5–6

Variable identification and fair-test design. “The steeper the ramp, the further the ball travels.” Friction step: “But the surface the ball rolls on might change the distance more than the angle.” Students name which variable they will control and which they are testing.

Implementation

Decision Checkpoint
The friction step and commitment are completed after the prediction is written and before any investigation or equipment collection begins. If a student cannot name a friction point, the investigation may be proceeding from assumption rather than hypothesis.
Teacher Judgement Note
A friction point as vague as “my prediction might just be wrong” without a specific mechanism signals the class needs a modelled example before independent work.
Related Frameworks

Position First Protocol · Evaluation Gate

Tony Jones · Founder, Field-Based STEM · Tomorrow Ready Resources · Free to use and share