§ Diagnostic

Do you show automation bias?

Ten questions. Answer honestly — the score is only useful if the input is. You'll get a profile and a guardrail set tuned to it.

Question 1 of 10· 0%

~5 min left

Q 01
When an AI-generated report contradicts your gut, what do you usually do?
Trust the report — the model has seen more dataNote the gut feeling, then defer to the reportPause and reconcile both before decidingTrust the gut and dig into the report's assumptions
Q 02
How often do you ask to see the source data or citations behind an AI output?
Rarely — the summary is usually enoughOnly when it looks obviously wrongWhenever the decision has real consequencesEvery time, as a rule
Q 03
A vibe-coded prototype is presented in a strategy meeting. What tends to happen next?
We refine the prototype — it's a great starting pointWe debate features shown in the prototypeWe step back and re-open the problem statementWe treat it as one option among several unbuilt ones
Q 04
How does your team handle disagreements with an LLM's recommendation?
They usually don't — the model tends to winSenior voices override; juniors defer to the modelWe surface the disagreement and document both sidesWe require the model output to be defended by a human
Q 05
When you review AI-drafted work, roughly what % of the reasoning do you re-derive yourself?
Under 10% — I'm checking style, not logicAround a quarterAbout halfMost of it — I re-run the reasoning
Q 06
How well can your team explain how the AI systems they use actually work?
They can't, and don't feel they need toOne or two power users could, roughlyMost understand the mechanics at a working levelWe invest in AI literacy as a standing capability
Q 07
Which sentence sounds most like your last board or leadership pack?
The model recommends…Analysis suggests…We recommend, based on X evidence and Y assumption…We recommend X; we considered and rejected Y and Z because…
Q 08
How often does someone on your team push back on an AI output in a meeting?
I can't remember the last timeOccasionally, from the same one or two peopleRegularly — we've made it psychologically safeIt's an explicit expectation of the role
Q 09
Have you built any guardrails specifically for AI-assisted decisions?
Written protocols, red-team reviews, sign-off thresholds, decision logs.
No — we rely on people to use judgementInformal norms onlyYes, for high-stakes decisionsYes, with regular audit and review
Q 10
When was the last time you deliberately worked a problem without AI, to keep the muscle?
I don't — it would slow me downRarely, and usually by accidentWeekly, for at least one non-trivial taskDaily — the model comes in only after I've formed a view

Ready when you are

Answer 10 more to reveal.

Do you show automation bias?

When an AI-generated report contradicts your gut, what do you usually do?

How often do you ask to see the source data or citations behind an AI output?

A vibe-coded prototype is presented in a strategy meeting. What tends to happen next?

How does your team handle disagreements with an LLM's recommendation?

When you review AI-drafted work, roughly what % of the reasoning do you re-derive yourself?

How well can your team explain how the AI systems they use actually work?

Which sentence sounds most like your last board or leadership pack?

How often does someone on your team push back on an AI output in a meeting?

Have you built any guardrails specifically for AI-assisted decisions?

When was the last time you deliberately worked a problem without AI, to keep the muscle?