RamaOnHealthcare August 4, 2023
The Importance of AI Governance Boards
Having watched and participated in an outrageous number of discussions and presentations where Artificial Intelligence is the main topic, it has become apparent that the AI winter is certainly over, and a new dawn of embedded AI is upon us. Nonetheless, the art of separating hype from fact is challenging for all.
Mohan Nair, CEO of Emerge Inc.
Caryn Lusinchi, Founder of Andromeda AI Inc. www.aiauditor.com
This interview, and those that will follow, highlight the key issues not about AI, but rather AI governance. This is a growing area of attention within boards that are chartered to manage risk and opportunities.
So, just what should be the concern of a board regarding AI? Is it part of the audit committee preview, or should it be somewhere else? What are (or should be) the principles that underpin AI in companies? Many have an AI department within an organization, but previously, it was held within the hands of a few. Now, AI is ubiquitously apparent across organizations and within our daily lives.
I have invited a specialist Caryn Lusinchi who has gained expertise in the subject of AI governance and with whom I hope to gain a preliminary understanding for Boards as to how to manage AI. In open disclosure, I am collaborating with Caryn to help boards frame this AI governance opportunity.
Mohan Nair (MN): Welcome Caryn to RamaOnHealthcare. Let’s go right into it. What can boards do in the context of managing opportunities and risk with AI in the board room?
Caryn Lusinshi (CL): Start your four-step program today!
- Adopt proactive awareness and identification of AI risks (through the lens of competitive risks, operational risks, and technology risks).
- Understand AI risk management!
- Explore ways to utilize AI beneficially as a risk management tool to foster innovation and create value.
- Establish a Board of Directors (BOD) and Senior management responsibility. Set up a policy-based AI governance system and authorize in-house and/or 3rd party tools for continuous monitoring of AI performance AND risks.
Adopt proactive awareness and identification of AI risks (through the lens of competitive risks, operational risks, and technology risks).
The sooner a BOD understands the importance of balancing enterprise AI value creation with risk management, the faster they can recognize potential risks and put mechanisms in place to mitigate them.
…recognize potential risks and put mechanisms in place to mitigate them.
To get started, the World Economic Forum has published two essential Empowering AI Leadership & Oversight Toolkit PDFs. This edition is geared toward a Board of Directors audience, while this publication is tailored for C-Suite.
Mohan: Please tell us more about Risk Management. For instance, it is easy for boards to just say it is too early to pin down opportunities because AI/Machine Learning/Generative AI is just starting to take form. Why not wait? Is that a good strategy, given the time?
Caryn: I suspect boards are adopting a wait-and-see approach… not because they want to but because they have little to no access to in-house AI/ML/generative AI subject matter experts as new and tenured talent are being lost to the big tech or generative AI startup brain drain. Plus, it’s been a challenge to receive real-time knowledge since technology innovation is evolving at hyperspeed. The 2022 AI Governance Report revealed that 58% of organizations either have no AI expertise on their Boards or do not know what proficiency they may have. Ninety percent of organizations have no structures and processes in place to control the use of AI internally and in their supply chains. Finally, 59% of Board members are not aware of what AI-related regulations are being considered.
Ninety percent of organizations have no structures and processes in place to control the use of AI internally and in their supply chains.
The good news is that you are not alone. Amid the roadblocks, stay curious and knowledge-hungry. Learn what you can, and when you can. Slowly start putting risk management guardrails in place and remain vigilant and agile. This provides the means to pivot when technology advances or the regulation pendulum swings.
Mohan: From your experience in Europe, have data standards like GDPR and regulations assisted the foundations for work in AI to be somewhat organized already? What is going on in the USA to assist in the consistent use and deployment of Generative AI?
Caryn: Yes, I believe the General Data Protection Regulation (GDPR) paved the way for the current EU AI Act. GDPR is the leading standard for personal data protection against forms of AI and automated decision-making systems. I’ve also heard the EU AI Act is reviewing which lessons can be drawn from GDPR enforcement precedents when making final decisions on EU AI Act’s scope and obligations.
The United States, to date, has adopted soft law strategies (principles, frameworks) for generative AI (examples: NIST and AI Bill of Rights). It is difficult to govern, given that generative AI, by nature, can never be used consistently in deployment as it is based on a neural network architecture that is probabilistic (based on probability) and non-deterministic versus predictive. The output is not 100% reproducible.
Mohan: Boards I speak to are struggling with what Generative AI can do and how to govern it. What do you suggest they do? Where can they learn more about governance on these matters besides asking you and me or other people in the field?
Caryn: The Harvard Law School Forum on Corporate Governance recently published a timely June 2023 piece entitled, “Generative Artificial Intelligence and Corporate Boards: Cautions and Considerations,” which is an excellent primer regarding BOD’s good faith role and fiduciary duty oversight in assuring management retains control over generative AI systems. It also advises establishing/expanding internal audit committees, compensation committees, and risk and management committees to regularly investigate and review the extent to which generative AI may be leveraged for high-risk use cases (human resources/human capital, financial services/reporting, healthcare, and more).
Mohan: Boards have three strategies to govern. Ignore it until later. Highlight it as an opportunity and risk with action items. Lean in and capture an early adopter mindset. Which do you suggest, given your background?
Caryn: Highlight it as an opportunity AND risks with action items. In the process, remember to engage diverse, multi-stakeholders (shareholders, employees, customers, suppliers/vendors, communities). Ask how AI can serve their divergent interests for specific use cases to find common ground, build brand trust, and create long-term value creation.
…engage diverse, multi-stakeholders….
Mohan: Thank you Caryn. Much of what we have discussed is basic governance and gaining knowledge before acting upon the issues. Further interviews will bring out more details as we gain more from our audience in terms of what they want to read. Please tell us what you want to read.
Mohan Bio: Mohan is the CEO of Emerge Inc., a transformation/innovation company. He is 3x corporate executive; 3x startup executive with 3 exits; 3x author (Wiley & Sons); and is a sought-after keynote speaker on Corporate Innovation, Inner innovation in the world of AI. He is an Edmund Hillary Fellow (New Zealand) and a Tando Global Fellow. Find him at linked/in/mohanemerge. Consider mohannair.com to see more of his work.
Caryn’s Bio: Caryn Lusinchi is the founder of www.aiauditor.com, which offers AI strategy & governance consulting to global enterprises, startups, and government organizations. She holds FHCA (For Humanity Certified Auditor) credentials under the EU AI Act, GDPR, and NYC Bias Law and has a certificate in Foundations of Independent Audit of AI Systems (FIAAIS). Previously, she led product marketing strategy & scalability for Arthur AI, which improved machine learning models for data accuracy, explainability, and bias detection. Additionally, she brings 15+ years of combined global enterprise expertise in B-to-C and B-to-B technology (ex-Google, ex-Meta), conversational AI, and product GTM roles across NA, EMEA, LATAM, and APAC.