#11 AI Oversight of Third-Party Vendors is Key
Plus: Evolving C-Suite roles, AI as a new species, Stanford AI Index, Meta's Llama3, and more
👋 Welcome back to "AI Simplified for Leaders," your weekly digest aimed at making sense of artificial intelligence for business leaders, board members, and investors. I invite you to explore the past issues here.
In this issue, I cover:
AI news: Meta’s Llama 3, Coco-Cola’s $1.1bn AI commitment, 2024 AI Index Report from Stanford
The evolving roles of C-suite in the age of AI
AI is a new digital species: Mustafa Suleyman’s TED Talk
My piece published on FBV: AI Oversight of Third-Party Vendors is Key
A way to kickstart AI adoption at your organization
A reader poll
Enjoy.
AI News, Curated for You
In this section, I highlight only the most significant AI developments with long-lasting impact or strategic implications for business leaders.
1. Meta's Highly Anticipated Llama 3 Release
After much anticipation, Meta released its open-source model Llama 3 on April 18. This release features pre-trained and instruction-fine-tuned language models with 8B and 70B parameters, supporting a broad range of use cases. The upcoming 400B+ tier is expected to be a game-changer, as Jim Fan, Senior AI Researcher at NVIDIA, notes:
"The upcoming 400B+ tier will mark the watershed moment that the community gains open-weight access to a GPT-4-class model. It will change the calculus for many research efforts and grassroots startups."
You can try Llama 3 on its official site, a lightning-fast Groq-hosted version here, or, if you're tech-savvy, create a completely local copy of the model with Ollama.
The AI Model Arena: A High-Stakes Game
The state-of-the-art AI model arena has been a whirlwind of activity, with the crown for the best-performing open-source AI model changing hands from Databricks to Mistral to Meta in just a few weeks. We can expect more rotation and the potential launch of an even more powerful OpenAI GPT-5 in the coming months.
A recent TechCrunch article questions whether there are simply too many models. The commoditization of models makes Sakana AI's Lego-like approach, which I covered in issue #8, particularly interesting.
Trust and Safety: A Must-Have for AI Models
Notably, Meta's Llama 3 introduction included new trust and safety tools as part of its off-the-shelf solution for responsible AI capabilities. This trend is being echoed by cloud and infrastructure players, making trust and safety features a must-have for any AI model.
2. Coca-Cola made a $1.1 billion commitment to accelerate clouds and AI
Microsoft and Coca-Cola have announced a 5-year, $1.1 billion partnership to accelerate Coca-Cola's cloud and generative AI initiatives. This deepens their existing technology collaboration, allowing Coca-Cola to leverage the Microsoft Cloud and Azure OpenAI Service for innovative use cases across its business. For many companies of different sizes, we have passed the first stage of ‘if’ and entered the stage of ‘how’ when it comes to integrating AI.
3. Stanford HAI published 2024 AI Index Report, its annual study of trends in AI
The 2024 AI Index Report from the Stanford Institute for Human-Centered Artificial Intelligence (HAI) highlights AI's rapid progress and growing commercial dominance. AI models now outperform humans on many tasks, but still struggle with complex reasoning. Notably, the private sector is leading AI research, releasing over 3 times more notable models than academia in 2023.
However, the skyrocketing costs of training top AI models, now reaching hundreds of millions, pose challenges for businesses. The report also cautions about the lack of standardized safety evaluations for large language models, making it hard to assess risks.
Yet, generative AI funding surged almost 8-fold in 2023, as companies race to harness this transformative technology. Importantly, AI regulations are proliferating, reflecting growing awareness of its societal impact.
The report suggests AI can boost productivity and address skills gaps, but warns of performance issues without proper oversight. It also tracks AI's accelerating influence on scientific breakthroughs across industries.
The Evolving Roles of C-Suite Executives in the Age of AI
As AI transforms businesses across industries, the question of who should lead and oversee AI initiatives has become increasingly important. Two recent perspectives shed light on the evolving roles of C-suite executives in the age of AI.
The Case for a Chief AI Officer (CAIO)
Shelly Palmer, a professor and thought leader in AI and technologies, presents a balanced view of the pros and cons of having a dedicated Chief AI Officer (CAIO) in companies. This question also extends to corporate boards, where many directors have sought guidance on whether a specialized AI expert is necessary.
While the answer depends on the specific context, in most cases, all board directors should strive to develop "AI fluency" rather than relying on a single expert. By doing so, they can ask pertinent questions and effectively fulfill their duty of care.
When it comes to operational leadership, I advocate for a dedicated AI leader. Rather than simply giving this person the decision rights on all things AI, I suggest a "strategic coach" approach. This individual should serve as a facilitator and advisor to C-level execs, identifying opportunities, understanding constraints and risks, driving collaboration, and ensuring the successful implementation of AI initiatives across the organization.
CFOs and CDAOs: Key Players in AI Strategy Development
A recent Gartner survey shows that only around one-third of CFOs are involved in developing generative AI strategies for enterprises.
This ratio is too low. My view: CFOs have to be in the room where it happens 🎙.CFOs need to understand the strategic reasons behind AI projects from the beginning, provide inputs to help prioritize among the many potential directions, and start developing frameworks to forecast and measure return on investments.
Another C-suite role that might need to be involved sooner, depending on the industry, is the Chief Data and Analytics Officer (CDAO). As AI initiatives heavily rely on data and analytics, CDAOs can provide valuable insights and guidance in the development and implementation of AI strategies.
The age of AI is reshaping the roles and responsibilities of C-suite executives. By fostering AI fluency among board directors, appointing dedicated AI leaders as strategic coaches, and involving CFOs and CDAOs in AI strategy development, organizations can better navigate the challenges and opportunities presented by AI.
A New Metaphor for AI: Digital Species
Mustafa Suleyman, CEO of Microsoft AI, delivered a thought-provoking TED talk last week.
Calling AI a 'tool' increasingly feels defensive and inadequate. As AI continues to develop at breakneck speed, we need a new metaphor that captures its potential without evoking fears and opens our minds to new possibilities.
In his presentation, Suleyman proposed a new way of thinking about AI: as a new digital species. In Suleyman's words:
"I think AI should best be understood as something like a new digital species...I predict that we'll come to see them as digital companions, new partners in the journeys of all our lives...This is, in my view, the most accurate and most fundamentally honest way of describing what's actually coming."
This perspective encourages us to view AI not as a mere tool or technology, but as a new form of intelligence that will increasingly interact with and shape our lives.
Suleyman also emphasizes that AI is not something outside of our human story, but rather a reflection of humanity across time:
"Since the beginning of life on Earth, we've been evolving, changing and then creating everything around us in our human world today. And AI isn't something outside of this story. In fact, it's the very opposite. It's the whole of everything that we have created, distilled down into something that we can all interact with and benefit from. It's a reflection of humanity across time, and in this sense, it isn't a new species at all."
This view of AI as a culmination of human knowledge and creativity invites us to embrace AI as an extension of ourselves rather than a foreign entity.
I highly recommend watching the entire TED presentation to fully appreciate the compelling storytelling and deliberate flow of ideas in Suleyman's talk. As business leaders, it's essential to consider new perspectives on AI and how they might shape our understanding and approach to this transformative technology.
Director's Corner: AI Oversight of Third-Party Vendors is Key for Fund Boards
Recently I shared my view on the AI oversight of third-party vendors for investment fund boards on Fund Board Views, a premier publication focused on providing practical guidance and thought leadership for registered fund directors.
I include a list of AI risks and questions for directors to ask, key areas of consideration, and other practical advice in the article. Here is a summary and I’d love to hear your feedback on the full article on FBV.
As the use of AI by investment funds and their service providers grows, fund boards need to exercise robust oversight over the risks associated with these third-party AI systems. Boards should establish clear vendor selection guidelines that prioritize AI governance, risk management, and control capabilities.
Due diligence on AI vendors should encompass their policies, processes, and track records, including transparency practices, model validation methods, and staff expertise. Boards may consider developing AI-specific criteria for evaluating vendors.
Ongoing monitoring is critical. Boards should require regular reporting from advisers on vendor AI risk assessments and material system changes. The frequency and format of these reports should match the complexity and criticality of the AI services provided. Boards should also have access to independent audits or assessments of vendors' AI practices.
Understanding the fund's overall AI strategy and initiatives is crucial for directors to provide effective oversight. Regular briefings can help the board stay informed about the evolving AI landscape and its implications. Reviewing fund disclosures on AI usage can also help ensure transparency and alignment with investor expectations.
By setting a high bar for vendor oversight and modeling responsible AI governance, boards can contribute to a more trustworthy and sustainable AI ecosystem. This is especially important as AI becomes more prevalent in the investment management industry.
A Way to Kickstart AI Adoption
In previous newsletter issues, I discussed the typical AI adoption path for businesses of all sizes, starting with task automation around document processing and information extraction. If you're a business leader looking to jumpstart AI adoption and build trust among your employees in using AI to amplify their work, Copilot for Microsoft 365 is a popular choice among your peers.
You can find Microsoft’s step-by-step roadmap here.
I have a suggestion to make the experience more delightful and empowering: involve as many roles as possible in the Copilot launch. The less constrained you are by existing functional lines, the more surprising collaborative creativity will emerge.
By engaging a diverse group of employees from the outset, you can foster a sense of ownership and excitement around AI adoption. This approach not only helps build trust but also uncovers unique use cases and opportunities for collaboration that might otherwise be overlooked.
So, take the first step today and watch as your organization embraces the transformative power of AI.
One More Thing
I’ve tried publishing this newsletter on different days to see if it leads to a difference in open rate and engagement. I have to say that it is inconclusive from data. If you have some view on this, please respond to this poll to help me out:
Thank you for reading. If you haven’t done so, please follow or connect with me on LinkedIn. I would love to hear from you.
Enjoy your week.
Joyce Li
Joyce - you write really well. I enjoy reading your contents. It’s educational, not too advanced where I would need to re-read to understand, and not too elementary either. Playing with Llama 3 now 😁
Thank you for sharing! Great post!