AI Learning Curve Becoming Steeper For Corporate Directors

The rapid evolution of generative AI challenges the ability of corporate governance to understand the new technology and to take the steps that may be necessary to monitor the application of AI within the organization.

This ability of computer systems to quickly generate detailed text and visuals from simple input prompts offers companies a wealth of exciting opportunities, as well as certain unknowns. It’s ultimately the board’s job to keep tabs on both. But it can’t do so without some basic familiarity with the systems.

As The Wall Street Journal observed, generative technology has the capacity to “insinuate technology more deeply into the way we live and work than it already is—not just answering our questions but writing our memos and speeches or even producing poetry and art.” It represents an exciting technological leap forward.

Indeed, that began with the seemingly innocuous introduction of Open AI’s ChatGPT application, which has since cascaded into a rapid-fire series of new AI product introductions with the potential to impact a breathtaking range of personal and business activities.

For example, Microsoft MSFT recently introduced an updated version of Bing, its AI driven search engine, that is powered by artificial intelligence software from OpenAI. Google GOOG has identified as a top priority the release of its own chatbot product, to be called Bard, which reportedly has been under development for months. In addition, the prominent Chinese technology firm Baidu BIDU is preparing to introduce its own chatbot, named Ernie (an abbreviated reference to “Enhanced Representation through Knowledge Integration”).

And other tech companies such as Jasper and Neeva are rushing to introduce their own variations of generative AI to the market with the promise of both enhancing search efforts and supporting traditional business functions such as finance, operations, legal, logistics and creative services. The perception is that generative AI is different than recent, previous highly publicized tech developments such as the metaverse, because it is grounded in years of existing matching learning which is already incorporated into many business protocols.

And as a result, this rapid introduction of generative AI has drawn significant attention from corporate boards, many of which have quickly recognized the potential benefits it may offer to their company’s operations through improved productivity and greater predictability.

But that attention must be balanced with a recognition of the significant learning curve that must accompany board oversight of corporate AI applications. Directors may indeed leave the details of implementation to corporate management, but they remain responsible for monitoring that implementation. As difficult and complex as the technical principles may be, boards must increase their related engagement with the principles of generative AI if they are to fulfill those responsibilities.

This enhanced engagement requires not only board education about the technology, but also an awareness of how generative AI may be applied by the company, and the social, legal, ethical and governance issues its use presents. These include the trust-related concerns of both consumers and employees; the necessary impact on board composition, including the possible need for subject matter expertise; the inevitability of government regulation; and the need to address the potential for “capability overhang.” All in all, it’s a long list.

Capability overhang refers to the hidden capacities of AI: skills and aptitudes latent within systems that researchers haven’t even begun to investigate. The specific concern is that generative AI models are so huge and complex-and the techniques available for evaluating them are still so juvenile-that it is difficult to fully understand both the opportunities that they offer and the risks that they present. And the potential for capability oversight may be exacerbated by the current explosion of AI image generators.

As tech commentator James Vincent notes, capability overhang isn’t necessarily a negative condition. It has the potential for concealing exciting additional capabilities, not just risks and threats. But by its very nature, it encapsulates the AI oversight challenge for corporate boards. In order for the board to monitor the organizational use of AI, and to make related acquisition and investment decisions, it must possess some basic awareness of the technological framework of AI and its intended organizational use. It also must be aware of the potential that the technology may offer unknown capabilities as well as unknown risks.

These are challenges that simply can’t be punted to management. The fundamental complexity of generative AI can’t be an excuse for the board to be excessively deferential to management on its organizational use. The extraordinary opportunities afforded by generative AI don’t allow the board to “go easy” on monitoring its application.

If the organization is going to make a commitment to implementing AI—as many will be doing—then the board must similarly make a special commitment to technical education and training, as well as to receiving dedicated management and consultant support to assist it in those efforts. As the consultancy Korn Ferry KFY has noted, “[t]oo many directors are not up to speed on AI, and made their mark in business in a ‘non-AI’ age.”

Generative AI is an exciting development with exceptional promise for commercial (as well as consumer) use. Yet related investment decisions are likely to be consequential for the company. Governing boards would do well to prepare themselves now, in order that their related decision-making and subsequent oversight be as informed as possible.

While that’s not a small task, it may nevertheless be an exciting one.

Michael wishes to acknowledge the assistance of his partners Jennifer Mikulina and Jed Gordon in the preparation of this post.