Decision-support screening · Bridge engineering

The task-based governance framework, put into practice.

The Bridge AI Risk Navigator is a screening aid for bridge managers and leadership. Describe the task, the data involved, and where the tool runs, and it returns a risk level with the review and safeguards that use calls for. Everything stays in your browser; nothing is sent or stored.

Assessment

Describe the intended use

Select one or more task frequencies to begin.

01 Professional context

These details tailor the review, responsible-charge, and training safeguards. They do not make a safety-critical task lower risk.

02 AI environment and governance

Deployment and data classification affect the chance that proprietary, controlled, or restricted information is exposed or handled outside approved conditions.

03 Bridge tasks and use frequency

Set the expected frequency for every task that the proposed use touches. Consequence and unreliability likelihood are expert-set screening values, not measures of a specific model's accuracy.

Tier 1 · language & knowledge Tier 2 · code & data Tier 3 · quantitative & structural
Method and limits

A transparent screening model—not a prediction.

The score identifies where consequence, frequency, deployment, data sensitivity, and governance gaps warrant stronger controls. It is an expert-judgment framework developed for discussion and local tailoring; it is not calibrated from the survey responses and does not validate a tool, a person, or an organization.

View scoring logic and thresholds

Inherent task risk: IR = L × C, with L = likelihood of an unreliable output and C = consequence if an error is not detected (each 1–3). Each selected task receives its own 0–70 task-exposure value based on its inherent risk and expected frequency. The tool retains the highest task-specific exposure, so adding lower-risk work cannot reduce the result. Deployment and data add 0–20 points. Governance status adds 0–10 points.

Screening levels and gates: Routine 0–19; Guarded 20–44; Elevated 45–69; Critical 70–100. Restricted information, controlled information in a consumer or personal service, and Tier 3 work in a consumer or personal service trigger a primary Not eligible to proceed status; any numeric score is supplementary. A Tier 3 task always triggers independent verification and PE responsible-charge safeguards.

Use and human factors: intended use routes review and documentation safeguards for exploration, internal technical decision support, or a project decision or issued deliverable; it does not add numerical points. Role, experience, licensure, and organization determine appropriate supervision, training, review, and contracting safeguards. They do not reduce inherent task exposure or replace professional judgment.