Psychological Safety Diagnostic

A set of observable behaviors and team signals that indicate whether psychological safety is sufficient for AI-assisted development. Helps Scrum Masters assess the team's safety baseline before introducing accelerated AI workflows.

Syntax

scrum-master-ai

Psychological Safety Indicators (AI-Assisted Team Context)

High Safety Signals:
- Developers disclose AI tool failures in standup without prompting
- Team members say "I don't understand this well enough to approve it"
- Retrospective surfaces quality concerns about AI-generated code
- Estimation includes honest discussion of AI contribution uncertainty
- Junior developers speak as frequently as senior developers in ceremonies

Low Safety Signals:
- PRs approved within minutes of submission consistently
- No one ever mentions AI getting something wrong
- Retrospectives produce the same action items repeatedly without follow-through
- Junior developers consistently defer to senior developers on AI code decisions
- "Looks good" is the most common PR comment on AI-generated code

Example

scrum-master-ai

// Scrum Master observation log after two AI-assisted sprints:

observations: {
  standupAiDisclosure: "Zero instances of anyone mentioning AI failure in two sprints",
  prReviewTime: "Average 4 minutes for AI-heavy PRs — likely not reading the code",
  retroThemes: "Consistent: improve code quality. No one names AI as a variable.",
  juniorDeveloperVoice: "Two junior devs have not spoken in retrospective in 3 sprints"
}

assessment: "Low psychological safety for AI-specific quality conversations"

interventions: [
  "Open next standup with: I noticed something interesting in our PR reviews this sprint — let's talk about it",
  "Explicitly celebrate the next instance of AI failure disclosure",
  "Add DoD agenda item: did anyone approve something they didn't fully understand?"
]