3#6 - Ieva Martinkenaite - Responsible AI: Ethical Governance and Positive Social Impact (Eng)

MetaDAMA - Data Management in the Nordics

«A combination of strong buy in from top-management and strong flow of change agents (…) is a requirement for succeeding.»
Eager to unlock the secrets behind building a trustful relationship with AI systems? I am sitting down with Ieva Martinkenaite, head of Telenor's Research and Innovation department to shed light on the interplay of accountability, ethics and AI technology. Through her role as translator between tech, leadership and business , Ieva brings a refreshing vantage point to the dialogue, providing a unique bridge between the tech and business spheres.
We're taking a deep dive into the creation of responsible AI within an organization. Our conversation explores the firm foundation of clear values and top management's proclamations, to cultivate a bottom-up process for a governance structure. Understand the three-layer structure of AI governance and the imperative of expert support for data professionals. Plus, we'll be scrutinizing how adopting responsible AI as a core principle can fetch a positive social impact.
In the finale of our discussion, we underscore the essence of responsible AI use and the value of investing in data professionals. Discover how individuals and companies can not only fulfill, but surpass compliance standards. Remember, it's not just about employing AI responsibly but about finding a responsible approach that fits you as an individual and your company.
Here are my key takeaways:
The two scenarios of concern with AI in Telecom:

1. Missing out!
2. Messing up!

• Telecom still needs to catch up, but with a string focus on using and scaling AI technology.
• The biggest differentiator in the sector is applying methods and technology to provide the best customer service.
• To scale AI, you need to have very solid data capabilities .
  • Cloud native data platform with various continuously upgrading technologies.
  • Efficient and scalable storage and processing capabilities.
  • Data Governance structures to ensure accessibility and use of data in a secure, privacy friendly, ethical way.
• You need that foundation before you can start building advanced AI capabilities.
• Apart from data you need people who are data literate and technical adverse.
• A strong data culture is important, not just for the data experts in your organization, but for everyone.

Responsible AI

• ResponsibleAI should be build on a solid Data Governance foundation.
• The biggest concern of executives with AI is the lack of traceability with data.
• We need to a) understand what are the risks, b) create responsibleAI by design.
• Executive support and belief in the AI journey is key.
• Data professionals have a responsibility to communicate complexity, translate and apply their knowledge to ensure a more general data literacy.
• You should do anything possible to be able to explain how your models work.
• You need to ensure that it is save to talk about, also not understanding systems.

Steps to building Responsible AI Governance:

1. Decide as an organization on your core principles / value - how may they be challenged by AI?
2. Define your principles / values for AI - these should be AI specific, but adopted to your setting, concerning risk, portfolio, etc.
3. Make these principles / values actionable.
4. Seek endorsement
5. Build a Governance structure
6. Ensure training and awareness

Positive Social Impact

• Companies should feel a social responsibility to go beyond what is required to build better, more ethical systems and use of those.
• Ask yourself, why are you doing responsible AI and Governance? For compliance obligations or do you what to go beyond that to build based on high ethical standards?

Luvut

1. Introducing our guest Ieva Martinkenaite (00:00:00)

2. Responsible AI in the Telecom Sector (00:06:27)

3. Building Responsible AI (00:11:38)

4. Creating AI Governance Framework Steps (00:28:26)

5. Building an Ethical AI Governance Framework (00:40:20)

6. Responsible Use of AI and Data (00:46:20)