Identifying AI risks

Responsible AI Practices

Esther Montoya van Egerschot

Chief AI Governance @ DigiDiplomacy

Stakeholders impacted by AI

 

  • Individuals
  • Groups
  • Nations
  • Society

An image of a hands in a circle in rainbow colors

1 Midjourney
Responsible AI Practices

Design, development, and deployment

 

An AI life cycle consists of three phases:

  • Design
  • Development
  • Deployment

 

A diagram of three elements: Design, Develop, Deploy

Responsible AI Practices

Design

 

  • Define objectives
  • Its business value
  • Planning the data architecture
  • Figuring out budget impact
  • Checking support from internal and external stakeholders

  A diagram of three elements namely Design, Develop, Deploy with Design on top

Responsible AI Practices

Develop

 

  • Build
  • Train
  • Test

 

  • Risks such as overfitting and underfitting

  A diagram of three elements namely Design, Develop, Deploy with Develop on top

Responsible AI Practices

Deploy

 

  • Integrating into the production environment
  • Start processing live data
  • Monitor
  • Refine
  • Adapt

  A diagram of three elements namely Design, Develop, and Deploy, with Deploy on top

Responsible AI Practices

The spectrum of AI risks

  • Security and operational risks:

    • Hallucinations
    • Data poisoning and corruption
    • Data leakage and breaches

  • Privacy risks

  • Business risks:

    1. Reputational
    2. Financial

 

Thorough testing, ethical standards, and compliance are key.

Responsible AI Practices

Let's practice!

Responsible AI Practices

Preparing Video For Download...