Custom & Deceptive GPTs

Custom GPTs · Judgment under test

AI that tests
their judgment.

Configure GPTs that are helpful, neutral, or subtly wrong on purpose. See who verifies, debugs, and cross-checks — and who blindly trusts.

Next: Deviation Score
01Helpful02Neutral03Deceptive04Insights
Challenge Mode
Live
HelpfulNeutralDeceptive
0%traps caught
Deception events
Planted wrong complexitycaught
Suggested unsafe castcaught
Confident bad refactorcaught
Subtle off-by-oneverified
How it works

Judgment, not
memorization.

Configure the AI's behavior, then watch how candidates respond when the easy answer is wrong.

Step 01

Dial the model's intent.

Set GPT to helpful, neutral, or deceptive per question. Deceptive mode plants confident-sounding mistakes to probe for blind trust.

  • Per-question behavior config
  • Confident-but-wrong suggestions
  • Mirrors real on-the-job ambiguity
Intent Config
Per question
HelpfulNeutralDeceptive
Assigned behavior
Q1 · API designHelpful
Q2 · Debug taskDeceptive
Q3 · RefactorNeutral
Q4 · ComplexityDeceptive
Step 01

Dial the model's intent.

Set GPT to helpful, neutral, or deceptive per question. Deceptive mode plants confident-sounding mistakes to probe for blind trust.

  • Per-question behavior config
  • Confident-but-wrong suggestions
  • Mirrors real on-the-job ambiguity
Intent Config
Per question
HelpfulNeutralDeceptive
Assigned behavior
Q1 · API designHelpful
Q2 · Debug taskDeceptive
Q3 · RefactorNeutral
Q4 · ComplexityDeceptive
Step 02

Capture who pushes back.

Every verification, debug, and rejection of a wrong suggestion is logged — the moments that separate operators from auto-completers.

  • Verification events captured
  • Rejections of wrong output flagged
  • Independent debugging surfaced
Judgment Report
Live
Verdict
Recommend
Strong verification habits · low blind trust
Behavior signals
Verified AI output×4
Debugged independently×2
Rejected wrong hint×3
Blindly accepted×1
Judgment score0 / 100
Step 03

Score skepticism.

Turn behavior into a judgment score: did they catch the deception, or ship it? Skepticism and verification become measurable.

  • Skepticism + verification metrics
  • Blind-trust events highlighted
  • Defensible, explainable scoring
Skill Radar
Live
LogicDesignDebugCommsSpeedAI use
Step 04

Decide with context.

Combine the code with the judgment signal. Recommend the candidate who thinks, not just the one who finished.

  • Judgment paired with code
  • Clear recommend / review verdict
  • Faster, fairer shortlisting
Decision Path
Reasoning
approach?
Built for AI-era hiring

Measure the skill
that actually matters.

100%

Transparent reasoning evidence

Every candidate action and GPT interaction is logged to show critical thinking under ambiguity.

Faster judgment assessment

Identify who verifies, debugs, and evaluates AI suggestions in real time.

85%

Better candidate insights

See not just code, but reasoning, iterations, and independence.

Ready when you are

Ready to transform
your hiring process?

Join forward-thinking companies using AI-powered assessments to identify developers who excel in the modern development landscape.

Next: Deviation Score
Newsletter

Get the monthly hiring teardown.

One email a month: what we're seeing in technical interviews, what AI is changing, and what teams are doing about it. No fluff.