Managing the AI Frontier: Six Questions for Corporate Boards and Leaders to Ask in the Age of Artificial Intelligence

Recognizing that in today's rapidly evolving business landscape, understanding the intricacies of artificial intelligence (AI) is essential for corporate boards, our client AMJ Campbell asked us to lead a session with their board. Kudos to AMJ’s management for understanding that with AI's transformative potential permeating every industry, the board must grasp its nuances to harness opportunities for innovation while navigating the associated risks to safeguard their company's future success.

It's not just the board at AMJ who needs to understand how to assess their company’s AI readiness and progress based on the six questions we prepared for them. For that matter, it’s not just boards who need to be able to answer those questions; it’s all leaders who must manage a truly transformational business phenomenon.

You can view the presentation we delivered to the board, but be aware it is very high level, and most of the content was in the actual discussion. We've captured the topics we covered in detail below in this article.

To help prepare to answer the questions, we reviewed key trends and risks, we provided an overview of AI basics (notably the application of AI in the Microsoft ecosystem), and we defined some key terminology.

AI Trends and Statistics

There is almost no limit to the eye-popping statistics on AI’s reach and growth available through search and, well, AI. We chose five, courtesy of Forbes Advisor, that provide focused strategic perspective for business leaders, demonstrating that: the technology is no fad and is experiencing hyper growth; companies are viewing the productivity of AI as being corporate Ozempic; employees, on the other hand, are wary of job losses; and a small majority of clients are prepared to trust companies that deploy AI.

• The market size for AI is expected to reach $407 billion by 2027, up from $86.9 billion in 2022
• 64% of companies expect AI to increase productivity, with 25% turning to AI to address labour shortages
• 77% of people are concerned that AI will cause job loss in the next year
• 65% of consumers say they’ll still trust businesses who use AI
• 10% of vehicles will be driverless by 2030

These are critical considerations for corporate leaders who need to balance the needs of all their stakeholders, and consciously manage trade-offs when needed.

Why Risk Is On Our Minds

No discussion of AI is complete without an acknowledgement of widespread and justified awareness of social risks. The “Godfather of AI,” Geoffrey Hinton, has been especially eloquent on this topic.

Business leaders must remain up to date on those societal and macro concerns, while also proactively anticipating business-specific risks. There are many cautionary tales, such as a ruling by a Canadian tribunal that Air Canada must honour a discount promised by its AI chatbot, highlighting the legal implications of AI errors.

Incidents like these force business leaders to think about a set of risks, including:

• Missing out on a transformation cycle
  • Kodak has passed from being a poster child for how not to respond to digital transformation to providing an in-depth business case study for new generations of leaders
  • Based on these ample sets of lessons, few would consciously sit out a transformation cycle
  • Hesitancy to get started, however, could cause an accelerating competitive decline relative to early adopters
• Information accuracy aka “hallucinations”
  • Hallucinations are responses generated by artificial intelligence that are incorrect, misleading, or downright nonsensical — though the bots present them as fact
  • Even when protocols are in place to prevent it from happening, AI-powered chatbots hallucinate anywhere from 3% to 27% of the time
• Brand crisis/erosion
  • The Air Canada story demonstrates that AI-related brand fails provide easy media clickbait, which can result in a full scale brand crisis requiring time-consuming and expensive mediation
  • Less visible, but sometimes more damaging over time, are incremental erosions in customer experience and trust that can chip away at the brand equity that takes years to build… and rebuild
• Experimentation is too small or too large
  • There is simply no way to appreciate how to apply AI to one’s business without experimentation
  • When experimenting, failure is a desirable outcome because it provides learning and data. But, as the Air Canada story indicates, if the experimentation is too big it poses the risk of public issues
  • Likewise, if it’s too small, it won’t provide meaningful data
• Misaligned talent, training, management
  • It goes without saying that AI is only possible with technology
  • But successfully mastering the technology is only possible with the right people who have the requisite ability and training, and who receive the appropriate management and support

The good news for corporate boards and leaders is that they are only sitting in their (hybrid) chairs because they have proven their ability to assess the impact and probability that is inherent in all risk calculations. This assessment is unique to each organization, and once the leaders have the right information and data, they can put appropriate policies and guardrails in place.

The Four Types of AI

In order to manage the frontier of AI, business leaders must understand the four types of AI — reactive machines, limited memory machines, theory of mind, and self-awareness — as well as their respective implications. 

• Reactive Machines
  • These types of AI respond to stimuli in real-time without relying on past experiences or internal memory
  • They are programmed to execute specific tasks or behaviours based solely on the input they receive, making them efficient for tasks where immediate response is critical
• Limited Memory Machines
  • AI with limited memory utilize past experiences to inform their actions
  • They can analyze historical data to make predictions or decisions, enabling businesses to leverage insights from previous interactions or patterns to optimize processes, such as in customer service or supply chain management
  • This is the current state of ChatGPT and similar tools
• Theory of Mind
  • This advanced form of AI is capable of understanding and modelling the mental states of others
  • It can interpret human emotions, intentions, and beliefs, which is invaluable for applications like personalized marketing, customer relationship management, and human-like interaction in virtual assistants or chatbots
  • This is still a theoretical type
• Self-Aware
  • These AI systems possess a level of consciousness where they can comprehend their own existence and mental states
  • While currently more theoretical than practical, self-aware AI could potentially lead to significant advancements in autonomous systems, robotics, and decision-making processes
  • This would offer businesses unprecedented levels of adaptability and autonomy
  • Like theory of mind this is theoretical. It is also a cause for concern for many, with Terminator style scenarios achieving singularity where AI can self-improve well beyond humanity and decide humanity is a bigger liability than benefit

Why These Types of AI Matter to Business Leaders

• Enhanced Decision-Making
  • By leveraging limited memory machines, businesses can make data-driven decisions based on past experiences, leading to improved efficiency and effectiveness in operations
• Customer Experience Personalization
  • Theory of Mind AI enables businesses to better understand and anticipate customer needs, leading to more personalized and engaging interactions that enhance customer satisfaction and loyalty
• Competitive Advantage
  • Investing in AI technologies, including self-awareness research, can provide businesses with a competitive edge by enabling them to stay ahead of market trends, innovate faster, and adapt to changing environments more effectively
• Ethical Considerations
  • Understanding the capabilities and limitations of different types of AI is crucial for business leaders to navigate ethical dilemmas surrounding AI adoption, ensuring responsible use and mitigating potential risks to reputation and trust

Predictive vs Generative AI

Similarly, leaders must know the difference between Predictive and Generative AI.

Predictive AI leverages historical data to make informed predictions. It is often used in weather forecasting, flight pricing, and financial forecasting. Predictive AI can help businesses make data-driven decisions and anticipate future trends and events.

Generative AI, on the other hand, uses unsupervised learning to create new data based on an existing dataset. It is often used in writing, image, and speech synthesis. Generative AI can help businesses generate new content, designs, and ideas, and can also be used to enhance customer experiences.

Why it matters to a business leader:

• Predictive AI can help businesses make data-driven decisions and anticipate future trends and events
• Generative AI can help businesses generate new content, designs, and ideas, and can also be used to enhance customer experiences

Augmented, Assisted, and Autonomous Systems

• Augmented AI
  • Augmented AI refers to systems that enhance human decision-making by working collaboratively with humans to improve their capabilities
  • These systems leverage AI algorithms to provide insights, recommendations, or predictions, empowering human decision-makers with valuable information to make more informed and effective choices
  • They act as intelligent assistants, supplementing human expertise rather than replacing it entirely
• Assisted AI
  • Assisted AI automates simple tasks and supports human actions and decisions by providing automated tools or processes that streamline workflows and enhance productivity
  • These systems handle routine tasks, freeing up human resources to focus on more complex or strategic activities that require creativity and critical thinking
  • While Assisted AI can take action, it will always ask for human permission first
• Autonomous Systems
  • Autonomous systems are AI-driven entities capable of operating independently without direct human intervention
  • These systems can make decisions and take actions based on predefined rules or learning algorithms, adapting to changing environments and circumstances without human oversight
  • They range from self-driving vehicles and drones to fully automated manufacturing processes and smart infrastructure

Why These Types of Categories Matter to Business Leaders

• Improved Efficiency
  • Augmented AI, Assisted AI, and Autonomous Systems can significantly improve efficiency by automating tasks, reducing manual errors, and speeding up processes, leading to cost savings and increased productivity for businesses
• Enhanced Decision-Making
  • By leveraging these AI categories, business leaders can access real-time insights, predictive analytics, and automated decision-making capabilities
  • This enables them to make data-driven decisions with greater accuracy and confidence, thereby minimizing risks and maximizing opportunities
• Scalability and Adaptability
  • Augmented AI, Assisted AI, and Autonomous Systems offer scalability and adaptability, allowing businesses to respond quickly to changing market demands, scale operations efficiently, and stay competitive in dynamic environments
  • These technologies enable agility and flexibility, crucial traits for businesses seeking growth and innovation
• Risk
  • Assisted and Augmented allow the people in the organization mitigate the risk of incorrect AI
  • With Autonomous AI the organization needs to understand the probability and impact of errors in the AI process

Open AI

OpenAI is a leading organization dedicated to advancing the development and ethical implementation of AI. Founded in 2015 by prominent figures such as Elon Musk, Sam Altman, and others, with Microsoft as a major investor, OpenAI operates with a mission to promote the creation of friendly AI that benefits humanity. By fostering collaboration and research in AI ethics, safety, and transparency, OpenAI aims to mitigate potential risks associated with AI while ensuring that its benefits are distributed equitably across society.

Business leaders should follow OpenAI due to its influential role in shaping the trajectory of AI development and its commitment to prioritizing ethical considerations. By staying informed about OpenAI's research, initiatives, and recommendations, business leaders can gain valuable insights into emerging trends, best practices, and potential risks related to AI adoption.

Microsoft AI Tools

Microsoft is a critical player in AI, and has established itself as a thought leader that generously shares terrific resources on the technology. An important benefit for organizations managing AI in the Microsoft ecosystem is that they can run applications on their own data, and it remains private to themselves. When using the public ChatGPT service, any content you upload as part of your interactions has the potential to become part of the large language model, and could be returned to anyone else using the service globally. With Microsoft, your data stays private to you, including not being shared with Microsoft themselves.

For business leaders seeking to navigate the landscape of AI technologies, understanding Microsoft's offerings, particularly Copilot and Azure, is crucial. Copilot, an end-user tool, serves as an AI-powered conversational assistant capable of providing information and answers to users' inquiries. Beyond simple responses, Copilot can also analyze, summarize, or generate content, enhancing productivity and decision-making processes within organizations.

On the other hand, Azure offers a comprehensive suite of developer tools for building customized AI solutions. With Azure Open AI and Azure AI Search, developers can create their own search and AI experiences tailored to their specific business needs. These tools provide access to prebuilt and curated models from Open AI, Meta (formerly Facebook), and other leading organizations, enabling businesses to leverage state-of-the-art AI capabilities without the need for extensive expertise in AI development.

Business Leaders Need to be Aware of Copilot and Azure for Several Reasons

• Enhanced Productivity
  • Copilot's conversational AI capabilities empower end-users to access information and insights quickly and efficiently, streamlining workflows and improving productivity across various functions within the organization
• Customized AI Solutions
  • Azure's developer tools enable businesses to build tailored AI solutions that address their unique challenges and opportunities
  • By leveraging Azure Open AI and AI Search, organizations can create personalized search experiences and AI-driven applications that meet the specific needs of their customers and stakeholders
• Access to Leading AI Models
  • Through Azure's integration with Open AI and other top AI providers, businesses gain access to cutting-edge AI models and algorithms, accelerating the development and deployment of AI-powered solutions
  • This access allows organizations to stay at the forefront of AI innovation and maintain a competitive edge in their respective industries

Overall, being familiar with Copilot and Azure equips business leaders with the knowledge and tools necessary to harness the power of AI effectively, driving innovation, efficiency, and growth within their organizations.

Some Artificial Intelligence Terminology

We’re nearly ready to review the six questions boards should be asking management to manage the AI frontier, but first, it’s helpful to be familiar with key terminology.

• Chatbot
  • A chatbot is an AI-powered computer program designed to simulate conversation with human users, typically through text-based interfaces
  • Chatbots use natural language processing (NLP) and machine learning algorithms to understand user queries, provide relevant responses, and perform tasks such as customer service, information retrieval, and automated transactions
• Large Language Models
  • Large language models (LLMs) are advanced AI models capable of understanding and generating human-like text at scale
  • LLMs, such as OpenAI's GPT series, utilize deep learning techniques and vast amounts of training data to learn patterns in language and generate coherent, contextually relevant text across various applications, including natural language understanding, text generation, and language translation
• Deep Learning
  • Deep learning is a subset of machine learning that utilizes artificial neural networks with multiple layers (deep architectures) to learn complex patterns from data
  • Deep learning algorithms, inspired by the structure and function of the human brain, excel at tasks such as image and speech recognition, natural language processing, and autonomous decision-making, by automatically extracting hierarchical features from raw data
• Limited Memory
  • Limited memory in AI refers to systems that have constraints on the amount of historical data they can retain and utilize for decision-making
  • Unlike systems with unlimited memory, limited memory AI models, such as recurrent neural networks (RNNs) or memory-augmented networks, store and access only a subset of past information, making them suitable for sequential data processing tasks where memory resources are constrained
• Hallucinations
  • In the context of AI, hallucinations refer to erroneous or nonsensical outputs generated by machine learning models, particularly in generative tasks such as image or text generation
  • Hallucinations can occur when AI models extrapolate beyond their training data or encounter ambiguous inputs, leading to unexpected and often unrealistic outputs that do not align with the intended task
• Prompt Engineering
  • Prompt engineering involves designing or crafting input prompts to guide the behaviour of AI models and influence their output
  • By carefully crafting prompts tailored to specific tasks or objectives, prompt engineering can enhance the performance, interpretability, and control of AI models, particularly in settings where fine-tuning or customization is desired
• Grounding
  • Grounding in AI refers to the process of connecting abstract concepts or representations learned by AI systems to concrete, real-world experiences or observations
  • Grounding enables AI models to relate symbolic or conceptual knowledge to perceptual inputs, facilitating meaningful interactions with the environment and supporting tasks such as robotics, language understanding, and decision-making in complex domains

Six Questions For/From The Board

According to the Harvard Law School Forum on Corporate Governance:

Board responsibility for managing and directing the company’s affairs requires oversight of the exercise of authority delegated to management, including oversight of legal compliance and ethics and enterprise risk management. The board’s oversight obligations extend to the company’s use of AI, and the same fiduciary mindset and attention to internal controls and policies are necessary. Directors must understand how AI impacts the company and its obligations, opportunities, and risks, and apply the same general oversight approach as they apply to other management, compliance, risk, and disclosure topics.

It’s a well-established principle for leaders to keep teams accountable by asking smart questions that prompt reflection, encourage clarity of goals, and foster a culture of accountability and ownership. 

The following six questions will help boards understand how management is addressing AI impacts. And even if management doesn’t have a board asking them these questions, being able to answer them will be essential for managing any company’s AI frontier.

1. Do you understand the basics? LMM, ML, NLP, NLU, NLG, IoT, AI versus automation, and more. These acronyms, and the distinction between them, are extremely important. There are two components of a Natural Language Processing (NLP) system, for example: Natural Language Understanding (NLU) and Natural Language Generation (NLG). When you input a text into an NLP engine, the meaning or context of the user is deciphered by the NLU construct, and the response is generated by NLG. This formula makes chatbots work, but how many people stop at "do we have a chatbot?"
2. How are you assessing and supplementing AI-readiness? Make sure there is a formal plan in place to assess the organization's AI readiness, and that management has conducted a skill gap analysis across departments to identify areas lacking expertise in AI technologies and methodologies, while also evaluating the board's understanding of AI's strategic implications. Actions to improve AI readiness may include investing in employee training programs, hiring AI specialists, forming strategic partnerships with AI-focused firms, and fostering a culture that encourages experimentation and continuous learning in AI-related fields.
3. Who is systematically monitoring competitors and adjacent industries? In our experience, people are intrinsically motivated to stay on top of AI. The key issue at the board and leadership level is channelling that analysis into an analytical framework. Decide what developments they must track — with the threat of competitive disadvantages being paramount — and getting monthly or quarterly updates on critical and actionable developments. 
4. Are you future proofing the work? Data needs to be carefully cultivated in order to leverage it for decision-making and AI. Gathering the wrong data, or in the wrong manner, can restrict the potential future uses of the dataset, as well as expose the organization to increased legal and other risks. How is management overseeing the organization’s plans for gathering, organizing and securely storing data?
5. What can your people do better than machines? Humans often outperform AI when judgment and manual dexterity are involved due to their ability to adapt quickly to unpredictable situations, incorporate emotional intelligence into decision-making processes, and execute intricate tasks with precision and creativity that AI currently struggles to replicate. Additionally, human judgment can incorporate ethical considerations and contextual nuances that are challenging for AI algorithms to fully comprehend and address.
6. Are you using AI regularly? No, it’s not a trick question because the answer is that everyone who types a question into a chatbot or takes an Uber uses AI. Is the board really using it by, for example, recording meetings in Teams and then having Copilot summarize the meeting? Is management using predictive AI to analyze historical data on customer demand and market trends to forecast future demand, allowing for optimized inventory management and staffing levels in their operations? In all cases, are people pausing to think about the risks we identified above and seeing what their own experience shows them?