Prueba de gobernanza de la IA

Las respuestas y explicaciones aparecen a medida que realiza el cuestionario.

1. ¿Cuál es el propósito central de la gobernanza de la IA?
   • Choose one model architecture for every organization
   • Set roles, rules, and controls for who builds, evaluates, deploys, and responds to AI systems
   • Replace technical testing with public statements
   • Guarantee that deployment carries no risk
2. Un laboratorio desarrolla una sólida evaluación de seguridad, pero ninguna decisión de liberación la utiliza. ¿Qué brecha de gobernanza persiste?
   • The model has too many parameters
   • The evaluation must be replaced with a product survey
   • The evaluation lacks an effective process or authority that makes it consequential
   • The lab has already proven safe deployment
3. ¿Qué alcance del ciclo de vida refleja mejor el enfoque de gobernanza de la guía?
   • Only the moment a model is launched
   • Only the initial training run
   • Only public relations after an incident
   • Data rights, training transparency, pre-release tests, staged release, monitoring, and accountability
4. ¿Qué combinación distingue correctamente las herramientas de gobernanza?
   • Risk assessments are procedural; usage monitoring is technical
   • Rate limits are procedural; model cards are technical
   • Both model cards and rate limits are legal penalties
   • A kill switch is only a performance metric
5. ¿Qué puede pasar cuando las reglas carecen de pruebas mensurables?
   • They automatically improve model accuracy
   • They can become checkbox compliance without evidence of system behavior
   • They eliminate the need for accountable owners
   • They ensure every risk is detected
6. Why can a third-party audit be part of AI governance?
   • It guarantees that no failure will occur
   • It transfers every deployment decision to the auditor
   • It can provide an independent check on claims or controls
   • It replaces ongoing monitoring
7. Which action is an example of production governance after release?
   • Stop documenting model versions
   • Treat voluntary evaluations as mandatory law
   • Remove access controls after launch
   • Monitor use and establish incident reporting for serious harms
8. Why does the guide say procedural and technical controls need each other?
   • Controls without enforcement may be ignored, while rules without tests may be symbolic
   • Technical controls make institutions unnecessary
   • Procedures always slow systems without benefit
   • A policy document alone proves a system safe