Implementing strategies for RAI

Responsible AI Practices

Esther Montoya van Egerschot

Chief AI Governance @ DigiDiplomacy

8-Step journey to Responsible AI

Embrace AI governance
AI Playbook
Identify key internal and external stakeholders
Leverage internal support mechanisms
Embrace a multi-stakeholder approach for external stakeholders
Explore additional responsible behavior indicators
Implement governance tools
Monitor, audit, and evaluate AI governance

Step 1: Embrace AI governance

Organizational buy-in
Explaining why it matters
Leadership: it's business critical!

Misconception:

AI governance is often viewed merely as a legal or regulatory compliance issue, neglecting its broader role in ensuring ethical, transparent, and accountable AI practices.

Step 2: AI playbook

Putting it all down on paper
Strategic necessity
Aligning with business goals
Aligning with ethics
Ensuring regulatory compliance
Building trust

Misconception:

There's a belief that strict governance frameworks may hinder innovation by imposing too many rules. In reality, well-designed governance encourages responsible innovation.

Step 3: Identify key internal and external stakeholders

Internal stakeholders

External stakeholders:

Employees
Unions
Customers
Regulators
Community organizations

Misconception:

Engaging with stakeholders is sometimes seen as a box-ticking exercise, rather than an ongoing process critical for the adaptive and inclusive development of AI technologies.

Step 4: Leverage internal support mechanisms

Leverage internal support through Codes of Conduct, Ethics Boards, and AI Squads.

Support their AI Literacy!

Misconception:

Relying solely on ethics boards or committees for ethical AI oversight can overlook the importance of integrating ethical considerations into every stage of AI development.

Step 5: Embrace a multi-stakeholder approach

Colorful image of crowded office with round table sessions

Misconception:

Involving external stakeholders can slow down the AI development process by introducing too many competing interests and opinions.

In reality, this inclusive approach enriches the process, making sure the AI is aligned with the needs and values of a broader community.

¹ DALL·E

Step 6: Explore additional responsible behavior indicators

A picture of the BCorp certification

Misconception:

There's a common misconception that only large corporations can afford to implement responsible AI practices, such as achieving BCorp status or engaging in ESG reporting, underestimating the impact and feasibility for organizations of all sizes.

¹ B Lab

Step 7: Implement governance tools

An image of a pyramid with the 3 layers

Misconception:

It's often believed that the same set of governance tools can be applied universally across all AI projects, ignoring the need for a tailored approach that considers the specific context and ethical implications of each project.

Step 8: Monitor, audit, and evaluate AI governance

Chief AI Governance Officer

Chief AI Officer (CAIO)

Misconception:

Once AI governance frameworks and policies are established, they do not need frequent review or updates.

This overlooks the dynamic nature of AI technology and societal values, which necessitate ongoing monitoring, auditing, and evaluation to ensure governance practices remain effective and relevant.

Let's practice!

Responsible AI Practices