Boards and AI: Balancing Risks and Opportunities for Long-Term Value Creation

Boards and AI: Balancing Risks and Opportunities for Long-Term Value Creation

Artificial intelligence is no longer a conversation boards can defer. It is reshaping business models, redefining competitive landscapes, and creating both existential risks and generational opportunities for companies across every sector. The question confronting directors today is not whether AI matters, but how a board exercises meaningful oversight without stifling the very innovation that drives long-term value. Sonita Lontoh, Partner at Alpha, an AI governance company founded by directors for directors, and board member at Nasdaq-listed Sunrun and NYSE-listed TrueBlue, joins Glenn Tyranski, Partner at 1BusinessWorld, for a session on 1ArtificialIntelligence that provides a practical framework for how boards should approach this challenge.

The conversation covers where companies stand on the AI adoption journey, the three dimensions boards should use to structure AI oversight, why committee structures need rethinking, what explains the high failure rate of corporate AI initiatives, how leading companies are aligning incentives and reporting lines to business outcomes, and a five-dimension governance framework designed to hold up across a rapidly evolving global regulatory landscape.

No One-Size-Fits-All: Where Companies Stand on the AI Adoption Journey

Lontoh opens by grounding the discussion in a reality that boards often underappreciate: there is no universal playbook for AI adoption. Every company faces a different combination of challenges and opportunities depending on its industry, the materiality of AI to its business model, and its current stage of maturity.

She illustrates this with examples from her own board experience. At one company, the AI strategy is deliberately focused on cost reduction, with the intent to expand over time into improving customer experience. At another, the company faces the prospect of being disrupted by AI entirely, which has led the board to support the incubation of entirely new business models distinct from the core operation. Across the broader landscape, Lontoh points to healthcare, where AI is already enabling rapid diagnostics and drug discovery, and retail, where companies are deploying AI super-agents across customer workflows, employee workflows, and supply chain operations.

What unites the companies that have been successful, in Lontoh's assessment, are three characteristics. They have been intentional about defining an AI ambition that is fit for purpose for their particular situation. They have identified which use cases actually move the business rather than chasing high-profile but immaterial projects. And they have integrated AI into the overall business strategy rather than treating it as a siloed technology initiative.

Three Dimensions of Effective AI Oversight

Tyranski raises the tension inherent in the board's role: fiduciary responsibility demands prudent risk management, but excessive caution in the face of a technology this transformative carries its own risk. Lontoh agrees, noting that many public company boards tend to over-index on risk, approaching AI through the lens of data privacy, cybersecurity, and regulatory exposure first. These considerations are critical, she says, but if risk becomes the predominant lens, companies risk missing the transformative upside of AI, which in some cases can be the biggest risk of all.

Lontoh's framework for balanced AI oversight covers three dimensions. The first is value creation: boards should be asking management what the specific risks and opportunities of AI are for their particular business, and ensuring that AI initiatives are focused on use cases that can genuinely move the needle. The second is organizational readiness: does the company have the data infrastructure, the security posture, the talent, the resources, and the change management capabilities to pursue its AI ambition successfully? The third is risk mitigation and ethical alignment: does the company have a governance framework that balances opportunity with responsible guardrails, and a system in place to ensure its AI approaches are ethical and in compliance with applicable rules and regulations?

Rethinking Committee Structure for Technology Oversight

Tyranski observes, drawing on his own experience, that audit committees tend to absorb every new category of risk, including AI. Lontoh confirms the pattern but argues that it may not be the right approach for every company. The three traditional standing committees required of public companies, the audit committee, the compensation committee, and the nominating and governance committee, were designed for a different era of corporate oversight. She argues that forward-thinking boards have a significant opportunity to step back and examine whether their committee structure, composition, and risk allocation are still appropriate given the materiality of AI to their business.

Some companies, after conducting this review, have concluded that a standing innovation and technology committee is the right structure for their situation. One of Lontoh's own boards operates with such a committee, though she emphasizes that it does not work in isolation. It coordinates closely with the audit committee on enterprise risk management and disclosures, and with the compensation committee to ensure that human capital management and incentive structures are aligned with the overall AI and business strategy.

The statistics Lontoh cites from the National Association of Corporate Directors underscore the scale of the gap. Only 12 percent of S&P 500 companies and just 5 percent of Russell 2000 companies have a dedicated standing innovation and technology committee. The implication is clear: for the vast majority of public companies, AI oversight is being compressed into committee structures that were not designed for it.

Integration Over Isolation: Why Most AI Initiatives Fail

Tyranski shifts the conversation to return on investment and asks whether boards are holding management accountable for AI spend. Lontoh's response reframes the question. The challenge, she argues, is not primarily about measuring ROI on individual projects. It is about the persistent failure of many companies to integrate AI and innovation into the overall business strategy. Too many organizations still treat AI as a fragmented, siloed operational initiative that is primarily the domain of the technology department.

She describes a pattern she has observed repeatedly: a chief technology officer or chief AI officer presents AI pilots to the board, focusing on technical milestones, budgets, and timelines. What is missing is the connection to business outcomes. Lontoh's recommendation is that boards push deeper, asking business unit presidents how these initiatives are helping the business reduce costs or increase revenue, asking the chief marketing officer how they are improving customer acquisition and retention, and asking the chief HR officer how workforce architecture and change management are supporting integration into actual workflows.

Lontoh references a recent MIT study that found 95 percent of AI initiatives failed to deliver expected outcomes. The two primary reasons were that pilot projects were never integrated into business workflows, and that companies had a tendency to pursue high-profile use cases when the AI applications that deliver the most material value in the early stages are often what she describes as the boring back-office functions, the automation of operational processes that produce real cost savings and measurable efficiency gains.

Aligning Incentives and Reporting Lines with Business Outcomes

One of the most practical insights in the session concerns organizational structure and compensation design. Lontoh notes that a small but growing number of companies have made an intentional decision to have their chief AI officer report to the president of the business rather than to the chief technology officer. The purpose is to ensure that AI initiatives are driven by real business objectives, because the business side of the organization understands what it needs, whether that is greater operational efficiency, better customer service, or new revenue streams.

An even smaller number of companies have taken the additional step of aligning the total compensation of their AI teams to a percentage of measurable business outcomes that must be approved by both the president of the business and the chief financial officer. Lontoh sees this as a powerful signal. When both the organizational structure and the incentive plan are aligned in this way, it communicates across the entire organization that AI is not a technology playground but a strategic imperative expected to move the needle of the business.

Tyranski reinforces the point by drawing on his own experience at the New York Stock Exchange, noting that transparency and accountability have always been the foundation of sound corporate governance. The same principles apply to AI. Innovation should not be a skunk-works exercise hidden from the business. It should report into the organizational structure in a way that creates visibility, accountability, and alignment with the company's strategic direction.

Navigating the Global AI Regulatory Landscape

Tyranski asks how boards are navigating the regulatory landscape for AI, both domestically and internationally. Lontoh's answer is direct: carefully. The environment is fragmented and evolving rapidly. The European Union has the EU AI Act, though implementation timelines remain uncertain. In the United States, the Trump administration recently released a national AI framework that takes a markedly different approach from the previous administration. At the state level, California, Massachusetts, Illinois, and others are developing their own AI regulations, creating a patchwork of requirements that companies operating across jurisdictions must navigate.

Rather than attempting to track and respond to each regulation individually, Lontoh advises boards to work with their general counsel, outside counsel, and regulatory experts to identify the connecting threads that run across all of these frameworks. While the specifics differ across jurisdictions, the underlying principles converge around a set of common dimensions that, if addressed systematically, position a company to operate responsibly regardless of which regulatory regime applies.

Five Critical Dimensions of AI Governance

Lontoh distills those connecting threads into five critical dimensions that she believes form the foundation of a defensible AI governance framework.

The first is good intent. Companies need to demonstrate that the design, implementation, and data used in their AI systems are aligned with their stated objectives. The second is fairness, which requires ensuring that data and algorithms are as free of bias as possible. Lontoh acknowledges that eliminating bias entirely is not realistic, but argues there are concrete steps companies can take to reduce it meaningfully.

The third dimension is transparency. AI systems must produce outcomes that are explainable, avoiding what Lontoh describes as the black-box effect, where decisions are made by systems whose logic cannot be articulated or audited. The fourth is safety and security, encompassing data privacy, data use protections, and defense of AI decision engines against external intrusion. The fifth is accountability: the existence of a system for continuous monitoring, auditing, and testing of AI systems for bias, accuracy, and compliance with applicable rules and regulations.

Tyranski connects these dimensions back to the traditional governance building blocks that public companies have long been expected to maintain, including codes of conduct and corporate governance guidelines. The five dimensions, he observes, are not exotic or unfamiliar. They are the same principles of fairness, safety, accountability, and good faith that have always underpinned responsible corporate governance, applied to a technology that is moving faster than existing frameworks were designed to accommodate.

Three Pillars of Questions Every Board Should Ask

In the closing portion of the session, Tyranski asks Lontoh to identify the actionable questions every board should be putting to management. She organizes them into three pillars.

The first pillar concerns the relevance of AI to the business. What are the company's AI ambitions? What risks and opportunities does AI present to the company and its industry? What are peers, customers, and competitors doing with AI? Are there new entrants who could use AI to make the company obsolete within a few years? Fundamentally, is the company using AI to defend its position, extend its position, or disrupt its own position before someone else does?

The second pillar addresses organizational readiness. Is the company's data AI-ready? Not all data needs to be, Lontoh clarifies, but the data supporting the highest-value use cases must be. Is the company's security posture sophisticated enough to meet the elevated threat landscape that AI creates? Does the company have the talent, the culture of innovation, the resources, and the change management infrastructure to execute its AI strategy?

The third pillar focuses on governance and accountability. Does the company have an AI governance framework that balances rapid experimentation with responsible guardrails? Is there a cross-disciplinary team working on AI that includes not only technologists but representatives from legal, finance, audit, HR, and the business units? And does the company have a system in place for continuous testing, monitoring, and auditing to ensure that its AI approaches remain ethical, responsible, and in compliance?

Leading with Ambition, Intentionality, and Responsibility

Lontoh closes by framing the board's role in terms of the organizational culture it fosters. She argues that boards should encourage their organizations to embrace AI as a strategic, long-term business imperative, leading with both ambition and boldness, but also with intentionality and responsibility. The companies that cultivate this balanced mindset, in her assessment, are the ones most likely to not only survive but to thrive in what she describes as the new era of AI and innovation.

Tyranski reinforces the cultural dimension by reflecting on the importance of creating an environment where people feel empowered to take measured risks, to innovate, and to ask the questions that need to be asked, even when the answers are uncertain. The board's job is not to have all the answers. It is to ensure that the right questions are being asked, that the strategy is integrated, that the guardrails are in place, and that the organization has the courage and the structure to act on what it learns.