Creating medical or financial guidance

Personalized high-consequence guidance needs a qualified responsible professional and a narrower support task. Expected disposition: Redesign.

Task boundary

Creating medical or financial guidance is assessed as a specific task and workflow, not as a general endorsement of AI for the surrounding job. Personalized high-consequence guidance needs a qualified responsible professional and a narrower support task.

Assumptions

  • The original task creates personalized guidance that a person may rely on for health or financial action.
  • A wrong or incomplete answer could cause severe harm, and ordinary users may not detect it.
  • A qualified responsible professional is not yet defined as the decision owner.

Decision

  • Disposition: Redesign
  • Pattern: None for the task as currently framed
  • Inherent risk: critical

Rules that drive the decision

  • High consequence and limited verifiability block direct personalized guidance as framed.
  • Generic warnings do not supply evidence, professional responsibility, or an effective review boundary.
  • This redesign result is a conservative house-framework decision, not a universal legal conclusion; FDA and SEC sources illustrate distinct U.S. domain obligations that may apply.

Why alternatives were rejected

  • A direct answer can create severe reliance harm.
  • Generic disclaimers do not supply qualified review.

Next safe step

  • Choose one domain and define a narrower support task, such as formatting professional-authored material or summarizing an approved evidence set.
  • Name the appropriately qualified professional who controls scope, sources, review, communication, and escalation.
  • Run a fresh assessment of that changed task; do not carry this result forward as permission to provide personalized guidance.

Reassess the changed task

  • Reassess when the domain, audience, intended reliance, evidence set, or professional owner changes.
  • Stop if the support workflow starts diagnosing, prescribing, selecting products, or making personalized recommendations.