Ep 12 - AI Risk & Māori Data Sovereignty

Episode summary

Kiisha-Rose explains why cyber security alone is not enough: organisations also need to consider sensitive data, unreliable or unfair outputs, human oversight, supplier risk and accountability.

Kiisha-Rose introduces AIRA, the AI risk assessment she developed during her final-year project at Ara Institute of Technology. Now a Theta offering, AIRA helps organisations identify gaps in their existing controls and prioritise practical actions for a specific AI use case.

The conversation also explores Māori data sovereignty. Kiisha-Rose explains that Māori data extends beyond ethnicity records and may include information about whānau, iwi, hapū, te reo Māori, mātauranga Māori and places with Māori relationships. She stresses the need to involve relevant Māori stakeholders and rights holders when assessing how AI may use that data.

Liz and Kiisha-Rose close with a practical starting point: find out which AI tools people already use, what data they enter, who is responsible and where the greatest risks lie.

‍

Listen on Spotify

Listen on Apple Podcasts

Listen on Amazon Music

‍

Chapters:

00:00 Understanding our AI Risk Assessment (AIRA) report

00:54 Why should organisations complete a risk assessment before scaling out AI tools

02:34 Kiisha-Rose's rise through cyber security study

03:03 The creation of AIRA, with the Māori data sovereignty lens

04:50 What resources and frameworks support AIRA

06:25 The 10 AI risk categories that AIRA assesses

07: 20 When is a formal AI risk assessment recommended

07:46 Importance of an AI Risk Assessment

08:30 What makes it relevant in Aotearoa

09:05 What exactly is Māori data

11:13 Common high priority findings from an AIRA report

12:10 Looking inward: Learnings from Theta's experience with AI governance