Nick Curum Explains How to Operationalize AI Governance Reports

In a recent LinkedIn post, Nick Curum discusses a common pitfall in how organizations approach governance reports, particularly concerning Artificial Intelligence. Curum argues that many leaders fail to translate the insights from these reports into actionable operational changes, rendering them largely ineffective.

Curum highlights a prevalent issue: the passive consumption of governance documents. He states:

“Most people use governance reports the wrong way.

They read them.
Highlight a few lines.
Maybe mention them in a deck.

But nothing operational changes.”

This observation sets the stage for Curum’s exploration of how to make governance reports, specifically for AI, a practical tool for strategic oversight rather than a mere compliance exercise.

Shifting from Passive Awareness to Active Oversight

Curum elaborates on the transformation needed in board-level engagement with AI. Drawing on insights from Deloitte’s AI Board Governance Roadmap, he emphasizes the necessity for boards to transition from a state of ‘passive awareness’ to ‘active oversight’.

According to Curum, this strategic shift has significant practical implications. He points out that effective AI governance requires:

• A comprehensive inventory of all AI systems in use.
• Clear understanding of how each system impacts strategy and risk.
• Sufficient AI fluency among board members to ask pertinent questions.

Curum posits that this forms the foundational principle for effective AI governance, stating:

“A board cannot govern what it has not mapped.”

This principle underscores the idea that without a clear understanding of the AI landscape within an organization, meaningful governance is impossible.

Distinguishing Accountability, Responsibility, and Control

Further refining the approach to AI governance, Curum references EC-Council’s work on board-level AI accountability. He finds their distinction between accountability, responsibility, and control particularly useful.

Curum argues that boards should focus on governing outcomes and consequences, rather than the intricate details of AI models themselves. He explains:

“Boards govern outcomes and consequences.
They do not manage models.”

This distinction is crucial for developing practical governance tools. Curum notes that the AI Decision Register he developed is structured around fields such as ‘Accountable Owner’, ‘Human Escalation Point’, and ‘RAG Status’. He asserts that these are not merely administrative details but serve as critical ‘leading indicators’ of whether governance is genuinely effective or merely superficial.

Operationalizing Governance for Immediate Use

The core of Curum’s message is the practical application of governance insights. He contrasts the passive reading of reports with the active use of tools that can be immediately leveraged in business settings.

As Curum articulates, the goal is to move beyond theoretical understanding:

“That is the difference between reading a report
and turning it into something a CTO, COO, or board can use in a meeting straight away.”

To facilitate this, Curum offers his AI Decision Register, described as a ready-to-use workbook with three tabs, designed for immediate implementation by CTOs, COOs, and boards. This tool aims to bridge the gap between understanding governance principles and actively applying them to manage AI risks and strategies effectively.