Nov. 29, 2023
Great governance holds the key to “good” AI
In assessing AI’s data and functionality and its intersection with current and future regulatory compliance, it is important for boards and management to identify the key company ethics that can serve as a foundation for documentation, future analysis, and risk mitigation strategies.
DR. T. ON SECURITIES
The recent turmoil at OpenAI has made three themes very evident about the burgeoning field of artificial intelligence (AI): (1) governance is critically important to ensuring appropriate implementation and expansion of AI, (2) the organizational structure of for-profit coexisting alongside non-profit may not provide the best mechanism for ensuring the achievement of social good when it comes to AI, and (3) boards and management both have critical functions in ensuring proper corporate stewardship and must work together using a consistent framework that can be applied to manage AI within their companies. With the expected change to be ushered in by the advancement of AI, and the dire consequences of getting it wrong, we do not have time to waste and must ensure our organizational structures are optimized for good corporate governance to effectively manage the inevitable growth of AI.
Governance is Key
Cutting-edge technology can often upend traditional corporate dynamics. This is predominately a result of lack of information, a need for raw talent, a fast-paced competitive environment and an uncertainty around best strategic direction. With so many moving parts and an inability to predict what is around the corner, the key tenets of good governance prove to be powerful corporate tools.
Good corporate governance generally boils down to three key components:
Participatory decision making where all directors and members of management can have an active voice in decisions being faced by the company (and this tenet goes hand in hand with having a diverse and inclusive board and management team so that all of the voices do not sound the same);
Accountability where each director and each member of management takes personal responsibility for the trust granted to them by the shareholders and expends the effort necessary to make the best possible decisions for the company; and
Transparency, which is critical both internally and externally as it ensures appropriate disclosure across the board regarding issues a company faces, possible solutions, risks and consequences.
Good corporate governance is the foundation and that must be strong to ensure the rest doesn't topple over.
Using For-Profit and Non-Profit Governance Systems Together = Mismatch
The fundamental underlying governance structures of for-profit and non-profit enterprises are different. A for-profit enterprise has a mission of making money and its key constituents in that quest are shareholders (i.e., investors). The for-profit enterprise is organized in a way that will maximize profit, which leads to organizational systems and structures that are geared for efficiency. Divisions and team organization are established in ways that maximize product or service development and revenue generation. Boards of for-profit enterprises are elected by the shareholders (and can be removed by shareholders), and they have a duty of loyalty and care to the shareholders of the companies in which they serve. Part of this duty of loyalty and care does include risk assessment and for a cutting-edge area like AI, social impact is certainly one component of that risk assessment.
A non-profit enterprise on the other hand has a mission of social good and its key constituents are donors who provide the funds for the non-profit enterprise to meet that mission of social good. Non-profits are organized in a way to maximize mission, and this leads to organizational systems and structures where efficiency is not top priority. Divisions and team organization tends to be focused on communities and there may be an academic or research aspect to their organization and function. Boards of non-profit enterprises are elected by other board members of the non-profit (and can only be removed by other board members). While there is some presence of accountability given the enormous amount of donor funds that these boards oversee, the impact of that accountability is often nebulous.
Given the extreme focus of government, lawmakers, communities, education institutions and companies of all kinds on the social impact of AI, it is no wonder that companies like OpenAI thought that a governance structure that mixes for-profit and non-profit would offer the best of both worlds, but recent events clearly demonstrate that the different organizational structures operate better independently than together. Also, given the vast amount of technological advancement expected to come from the progression of AI, one could argue that the for-profit model is much better equipped from a governance perspective to drive progress. With a strong board who is knowledgeable and equipped with the tools necessary to manage the governance challenges imposed by AI, a for-profit enterprise provides a more solid fundamental structure with more checks and balances. While profit will always be the driver in such an organization, mission if cultivated appropriately by the board of directors and management team can become a critical piece of the cultural infrastructure that will necessitate decision making that considers both profit and mission.
Boards and Management Need an AI Systems Framework to Make Consistent Decisions
Directors and officers are required to exercise loyalty and care to the companies they serve and the shareholders of those companies. Recent case law has reinforced that a director or officer fulfills these duties through appropriate oversight over company activities and risks. In particular, courts have emphasized the need to be focused on systemic and knowable risks and to institute effective disclosure mechanisms and compliance programs to avoid liability. As such, it is important for all boards to begin to develop fluency and visibility about AI, the regulatory impacts and the risks posed for their respective companies. To do so, boards can rely on a simple framework.
As discussed above, mission is a critical quality in the management of AI. In assessing AI's data and functionality and its intersection with current and future regulatory compliance, it is important for boards and management to identify the key company ethics that can serve as a foundation for documentation, future analysis, and risk mitigation strategies. Depending on the AI implemented, boards may place different levels of magnitude on each of these risks and identifying the magnitude of such risks for each specific company is critical to being able to provide proper oversight. These key ethical risks can easily be remembered by the acronym FAIR.
F - Fairness: Boards should consider whether the AI results in biased outcomes and scrutiny should focus on whether the AI is doing more harm than good. This is a key regulatory tenent but is also a critical focus of employees, consumers and shareholders. A focus on this component will ultimately lead to better products which will drive more profit.
A - Accountability: Boards must be focused on accountability, both in terms of developing systems to identify unintended outcomes of AI (such as bias) and processes to correct those unintended outcomes. Continuous review is critical as the outcomes of AI will change by its every function and new unintended outcomes may emerge over time. This component is all about the feedback loop and boards and management who are focused on accountability can learn quickly from mistakes, adapt and more readily implement strategy that moves the company in the right direction.
I - Inspection: Inspection is about transparency and disclosure. AI systems should be available for inspection and any outcomes produced should be explainable. This ties directly into the documentation component, which is another key premise of proposed and current regulation. Inspection also builds trust and respect from stakeholders, thereby reinforcing mission and enhancing profit.
R - Reliability: AI systems should be reliable, safe and secure. This helps to generate trust among stakeholders, both internal and external to the company. Reliability ensures a level of consistency and that is going to contribute to both mission and profit.
The FAIR framework provides boards with a simple but effective approach to providing the necessary risk oversight for AI in their companies. The FAIR framework also allows companies to integrate mission considerations at each level, thereby ensuring that while profit is the main goal, mission can become such an embedded cultural aspect of every company decision that it will become a clear component of the company's strategy.
When it comes to "good" AI, corporate governance is key
As AI increasingly becomes the backbone of many of our systems, it is more important than ever that companies, management teams and boards adapt and learn the tools needed to better position their companies for success. This will not only provide their companies with a strong foundation for growth, it will instill a greater sense of trust amongst key stakeholders. This is becoming ever more important as regulation of AI is beginning to take shape. Given the complexity and fast pace of innovation, AI is an area where it will benefit boards to be leaders and get ahead of regulation as stewards of their respective companies rather than race to catch up. Good corporate governance is a critical requirement in the AI race as it provides the structure necessary to ensure both profit and mission can be front and center.