MIS Quarterly Executive Volume 22 Issue 4 December 2023 How Boards of Directors Govern Artificial Intelligence Benjamin van Giffen Helmuth Ludwig Follow this and additional works at: https://aisel.aisnet.org/misqe Article 7 Recommended Citation Giffen, Benjamin van and Ludwig, Helmuth (2023) "How Boards of Directors Govern Artificial Intelligence," MIS Quarterly Executive: Vol. 22: Iss. 4, Article 7. Available at: https://aisel.aisnet.org/misqe/vol22/iss4/7 This material is brought to you by the AIS Journals at AIS Electronic Library (AISeL). It has been accepted for inclusion in MIS Quarterly Executive by an authorized administrator of AIS Electronic Library (AISEL). For more information, please contact elibrary@aisnet.org. SIM MIS MARTERLY XECUTIVE EXE WHERE IT LEADERS CONNECT 45 TM How Boards of Directors Govern Artificial Intelligence Artificial intelligence is top of mind, even for nontechnical business executives and board members. However, the majority of boards struggle to understand the implica- tions of Al for their businesses and their role in governing it. We describe how some boards are addressing AI and identify four groups of board-level Al governance issues. We provide examples of effective board-level Al governance practices for each group of issues and make recommendations for establishing board-level Al governance. ¹,2 Benjamin van Giffen University of St. Gallen (Switzerland) Helmuth Ludwig Southern Methodist University (U.S.) Artificial Intelligence in the Boardroom Artificial intelligence (AI) is top of mind, even for nontechnical business executives and board members. In 2019, more than 75% of executives had already voiced co cerns that their companies might become obsolete if they failed to use AI effectively.³ Fast forward to 2023 and the majority of boards of directors are still struggling to understand the implications of AI for their businesses and the role they have in governing it.4 AI, as a transformative technology, is reshaping the business landscape across many industries, generating new value-creation opportunities and redefining competitive dynamics. The broad adoption of AI requires capital-intensive investments that are often uncertain in nature and pose novel legal, ethical and reputational risks. However, Al, like other digital technologies, is often not a mainstream boardroom topic.5 Boards, as stewards of shareholders' interests, have the fiduciary responsibility to oversee and direct the affairs of the company. Their mandate includes strategy approval, financial oversight, assurance of adequate enterprise risk management and executive appointments. 1 Martin Mocker is the accepting senior editor for this article. 2 The authors thank Martin Mocker, Blaize Horner Reich, Joe Peppard and the members of the review team for their thoughtful feedback and guidance throughout the review process. We also gratefully acknowledge the support of the interviewed board directors in providing insights into their AI governance practices. 3 For a discussion on the business criticality of AI technology perceived by C-level executives, see Reilly, A., Depa, J. and Doug- lass, G. AI: Built to Scale, Accenture, 2019, available at https://www.accenture.com/content/dam/accenture/final/a-com-migration/ thought-leadership-assets/accenture-built-to-scale-pdf-report.pdf. 4 Watch Board Effectiveness: A Survey of the C-Suite report, PwC's Governance Insights Center. May 2023, available at https:// www.youtube.com/watch?v=SLqBZD5zp7E. DOI: 10.17705/2msqe.00085 5 The need to govern technology at the board level is not new and has also been suggested in prior studies. See, for example: 1) Nolan, R. and McFarlan, F. W. "Information Technology and the Board of Directors," Harvard Business Review (83:10), October 2005, p. 96; and 2) Weill, P., Apel, T., Woerner, S. L. and Banner, J. S. “It Pays to Have a Digitally Savvy Board," MIT Sloan Man- agement Review (60:3), March 2019, pp. 41-45. December 2023 (22:4) MIS Quarterly Executive 251 252 Given Al's transformative impact, board- level responsibilities extend to overseeing the company's use of AI to fulfill the boards fundamental duties of care and loyalty. In this article, we use the term board-level AI governance, or Al governance for short, to refer to the way boards oversee and integrate the adoption and implications of Al within corporate governance practices. This article aims to help boards in establishing board-level Al governance. Driven by the question "How can boards effectively govern AI?", we conducted in-depth interviews with high-profile board members from diverse industries and with varying levels of digital and AI expertise (see the Appendix for details). Our observations reveal how boards are addressing AI, and provide insights into how boards are integrating AI into their corporate governance. Our analysis identified four groups of board- level Al governance issues: 1) Strategy and Firm Competitiveness; 2) Capital Allocation; 3) AI Risks; and 4) Technology Competence (see Figure 1). The issues in each group present specific challenges and opportunities, and boards should address them according to their organization's needs and priorities. AI, as a strategic enabler, can impact business strategy by introducing new competitors and business models. Capital allocation focuses on ensuring funding for Al experiments and strategically directing investments toward Al capabilities. Directors face new challenges in risk oversight due to Al's ethical, reputational and regulatory complexities. Lastly, enhancing technology competence within boards and executive teams ensures informed Al governance decisions. Below, we describe the issues of each group in detail, highlighting the associated challenges and opportunities, and report on the AI governance practices that were implemented within the boards of our respondents. We then provide three actionable recommendations to assist boards in their efforts to establish Al governance. 6 The duties of care and loyalty, alongside exercising business judg- ment, define the standards for directors' behavior and engagement as well as their legal liabilities. For more information, see Walker, G. and Madsen, T. L. Modern Competitive Strategy (4th edition), McGraw Hill, 2015, pp. 342-343. MIS Quarterly Executive | December 2023 (22:4) How Boards of Directors Govern Artificial Intelligence 1. Strategy and Firm Competitiveness Al Governance Issues AI is a strategic enabler and differentiator with the potential to deeply influence an organization's competitive position and bus model. In response, boards are adopting two specific practices: 1) ensuring that AI is reflected in business strategy-i.e., addressing how Al is impacting the strategic landscape and 2) incorporating AI into the board's annual strategy meeting, which ensures systematic board-level discussions of AI. Ensuring that Al Is Reflected in Business Strategy For many board members, AI is a new topic outside their experience and comfort zone and, as a result, boards may often view it primarily from a risk perspective. However, some boards are encouraging management to evaluate the opportunities of AI. Directors reported that their boards have reviewed, or plan to review, the potential for Al to create new offerings, evaluated the emergence of new competitors and assessed the impact of AI on their company's business model. Several boards have asked their executive teams to evaluate how AI is integrated into the strategy for unique market positioning through new Al-enabled offerings. Though approaches varied, there was often a particular focus on how AI can create customer value or cost advantages. An example of anchoring and communicating AI in a business strategy is provided by Ping An Insurance's latest business analysis, which details how it plans to empower financial services with technologies (e.g., by using an “AI Customer Visit Assistant") and how it strategically focuses on developing AI as a core technology to empower human service representatives with AI, offering services covering lending, credit cards and insurance.7 The board of another insurance company requested an analysis of the use of AI by competitors so it could evaluate the emergence 7 See Announcement of Audited Results for the Year Ended De- cember 31, 2022, Ping An Insurance Group, March 2023, pp. 51-55, available at https://group.pingan.com/resource/pingan/IR-Docs/2023/ pingan-ar22-report.pdf. misqe.org | © 2023 University of Minnesota Figure 1: Four Groups of Board-Level AI Governance Issues Challenges and Opportunities Al Governance Practices Strategy and Firm Competitiveness • Business strategy • New competitors • Business model changes • Ensuring that Al is reflected in business strategy • Incorporating Al as a key topic in the board's annual strategy meeting Capital Allocation Board of Directors Executing fiduciary duties in the context of Al • Experiments • Platforms and tools • Partnering, M&A investments • Encouraging Al experiments in annual budgeting process Securing investments in foundational Al capabilities • Supporting external Al partnering and M&As of new competitors. A director of this company noted: "Al use cases aren't typically discussed at the board level, but many of our directors wanted to understand our Al approach versus that of our competitors. Especially because Al is such a competitive weapon in insurance, both in terms of attracting new customers and managing claims and identifying potential fraud.” How Boards of Directors Govern Artificial Intelligence The analysis enabled the board to review and evaluate the company's future market positioning, ensuring that it did not risk falling behind and identifying potential competitive advantages and investment opportunities. As a result, the company made significant Al investments with a transformative impact on its business operations.³ In addition, several directors reported that their boards plan to discuss with management how to anticipate and address Al-driven business model changes. These include the emergence of new, evolved and potentially disruptive business models enabled by AI offerings that serve customers differently, often in a highly personalized and automated manner, while also providing cost benefits. For example, Ant Group- 8 This insurance company currently has an annual investment of around $200 million in AI-related software. Al Risks • Ethical and reputational • Legal and regulatory • Inviting Al risk experts to present to the board or audit/risk committee • Including Al in ERM (Enterprise Risk Management Technology Competence • Board composition • Executive management team composition and structure • Developing board directors' Al competency • Reviewing board composition and expanding committee roles • Including Al competency in succession planning and CIO search criteria an affiliate of Alibaba Group-offers AntPay, a payment technology that uses AI to assess creditworthiness, enabling Ant Group to provide microloans to consumers and businesses often overlooked by traditional banks.º Another example is Siemens Mobility's offering for Renfe, the national railway company of Spain, which uses AI to accurately predict the number of train passengers. This Al-enabled intelligent mobility-as-a-service (MaaS) platform has changed Siemens Mobility's business model from selling trains to offering passenger transport capacity services and has enabled Renfe to improve its on-time performance and increase revenue by €156 million over five years, driven by a resulting shift from airline to rail travel.¹0 From our interviews, we found that many directors view understanding the role of AI for the company and engaging in decisions about Al's role in business strategy as an important board responsibility. These directors focused on both traditional AI, which primarily uses machine learning to learn patterns from data to make predictions, and generative AI, which can 9 For more information on Ant Group's AI use, see: Cho, Y. "How AI and vast data support Ant Group's financial empire," Nikkei Asia, November 2, 2020, available at https://asia.nikkei.com/Business/Fi- nance/How-Al-and-vast-data-support-Ant-Group-s-financial-empire. 10 See Siemens Mobility and Everis to Develop a Mobility as a Service (MaaS) Platform for Renfe in Spain, Siemens, 2021, available at https://press.siemens.com/global/en/pressrelease/siemens-mobility- and-everis-develop-mobility-service-maas-platform-renfe-spain. December 2023 (22:4) MIS Quarterly Executive 253 Table 1: Four Ways of Incorporating AI into the Board's Annual Strategy Meeting Leader for 60-minute session External industry expert Focus External View ● Trends and drivers of Al within the firm's industry • Trends and drivers of Al in adjacent industries Internal Examples of What Has and Has Not Worked ● Overview of Al initiatives and applications in use (or not) within the firm • Overview of the current and future Al investment portfolio Overview of the Competitive Landscape Competitor positioning review and evaluation Traditional competitors and new entrants exploiting Al How Boards of Directors Govern Artificial Intelligence CEO's Al Agenda Assessment of Al impact and opportunities within/beyond the business model • Approach to creating new business value and competitive advantage with Al generate new, unseen content such as images or human-like text. Board members should not rely on executive management to initiate these conversations. As one director put it: “In all the boards I am part of, the topic of Al was not opened by executive management." This quote, echoed by others, underscores the board's duty to start the AI conversation. The question is: How best to approach it? Full board meetings provide an opportunity for board members to receive updates on AI and review and discuss the firm's strategic plans that involve AI. These full board discussions can facilitate a shared understanding and support for strategic Al-related decisions, which is important because, as one director noted: "Ultimately, board decisions are [whole] board decisions and not decisions of individual directors." Full board discussions, however, are often constrained by the limited available time, preventing the necessary depth in the conversation. An alternative approach was adopted by the directors of a pharmaceutical company that was under pressure to drive rapid Al-enabled technological change. They established a sounding board where Al-savvy board members could meet with the CIO and other key executives. This group met at least twice a year for project- focused dialogues about the progress of AI initiatives. 254 MIS Quarterly Executive | December 2023 (22:4) CEO, COO/CTO, CIO CEO, chief strategy officer CEO Incorporating Al as a Key Topic in the Board's Annual Strategy Meeting The annual board strategy meeting is an excellent opportunity to deeply involve the board in strategic decisions related to AI. In this way, boards can ensure that AI is addressed and integrated into the company's strategic agenda. The directors we interviewed described four types of sessions, each lasting about 60 minutes, where AI had been incorporated into their boards' strategy meetings, focusing respectively on an external view, internal AI examples (good and bad), the competitive landscape and the CEO's AI agenda (see Table 1). External view: In some boards, an external expert outlined AI opportunities, challenges and industry-specific best practices and how these related to the company overseen by the board. These sessions included a review of nontraditional competitors (e.g., digital natives, startups) that may be at an early or even experimental stage but show significant growth in the company's market. This type of session provides insight into how AI could redefine the company's competitive space both within and outside its industry and can even identify potential investment areas or possible acquisition targets (merger and acquisition opportunities). As one director noted, an external view is particularly helpful because "most strategic disruption originates from outside an industry." Internal examples of what has and has not worked: In this type of session, executive management provides an overview of AI misqe.org | © 2023 University of Minnesota