The 14× Problem
Research Brief
thinkinglabs.academy/research
14
×
more likely to get the wrong answer when AI answers first
Shaw & Nave, 2026 · SSRN
What the research says

Your child doesn't use AI.
They yield to it.

A peer-reviewed study tested what happens when students get an answer from AI before working through a problem themselves. The results were not subtle.

When the AI gave a wrong answer, students had fourteen times lower odds of getting it right compared to working alone. Worse: their confidence went up at the same moment their accuracy collapsed.

45.8%
Correct working alone — no AI
71%
Correct when AI was right
31.5%
Correct when AI was wrong

This isn't carelessness. Researchers call it "cognitive surrender" — the moment a student stops building the answer and starts adopting the machine's judgment as their own.

"The machine did not merely move answers. It raised confidence even when it was wrong. That miscalibrated metacognition may be the most dangerous part of the whole arrangement."

— Dr. Colin Lewis, The One Percent Rule

Peer response
"I really like the idea of a tool-free first week Syd. We've spent a decade trying to make everything frictionless, only to find that friction is where the actual learning is stored. It moves the conversation from cheating to judgment."
— Dr. Colin Lewis, The One Percent Rule · 30,000+ subscribers

The good news

The same study identified "the Independents" — students who worked through problems before turning to AI. Under pressure, they kept their edge. That posture is trainable. Ownership before extension. Attempt before delegation. That's what Thinking Labs builds.

Six weeks to build the student who doesn't surrender.

Thinking Labs is a 6-week cohort for students ages 13 and up. Week 1 is completely tool-free by design. Students build five cognitive dimensions — independent reasoning, information evaluation, persistence, communication and defense, and AI relationship. The goal is not to block AI. It's to build the student who can wield it instead of yield to it.

6
weeks, live cohort
5
cognitive dimensions assessed
12
students max per cohort
20+
years in the classroom

Start here — it's free

Two versions. Parents see the big picture. Students go deep on their own thinking. Both about 10 minutes. Both free.

For Parents
Cognitive Gap Assessment
Answer questions about your child's thinking habits across 5 dimensions. Get a full score and interpretation report.
Take the Assessment →
cognitive-gap-assessment.vercel.app/register
For Students
Student Self-Assessment
Students answer for themselves. Same 5 dimensions. Builds self-awareness and gives a baseline before the program begins.
Take the Assessment →
cognitive-gap-assessment.vercel.app/student/register

"Your child can get any answer in seconds. I teach them what to do when the answer isn't enough."— Syd Malaxos, Founder