Hands-on learning aids GenAI governance

Audit committees’ own AI use can sharpen their oversight

Editor’s note: This article first appeared in NACD’s Directorship® Magazine and is reprinted with permission. Copyright © 2024 National Association of Corporate Directors. All rights reserved.

As boards and audit committees endeavor to get their arms around generative artificial intelligence (GenAI) governance responsibilities, gaining experience and skills by using GenAI to improve their own work as directors can be a foundational step toward more effective oversight.

One compelling question for audit committee members, given the vast amount of data they deal with as part of their oversight responsibilities, is: How can GenAI be harnessed to elevate our own performance? Below are three actionable suggestions that audit committee members can use to capitalize on GenAI’s potential.

Use GenAI to quickly summarize information

Often, audit committee members are not able to efficiently read the vast number of materials presented or study them in detail. We have found that even after directors have read lengthy documents, it would be valuable to have a summary and analysis of key points; to be clear, this summary would never replace reading the documents in full.

Consider, for example, the challenge of sifting through a 200-page Form 10-K with a 50-page management discussion and analysis (MD&A). Audit committee members could use GenAI to create a summary and analysis of these documents so they can concentrate on the most pertinent information.

Broaden your view of past performance

GenAI can also empower audit committee members to broaden their understanding of the company’s financial performance over time via a GenAI analysis of historical financial information. Having GenAI synthesize a few years of key documents, especially qualitative information that may not be easily captured in the numbers, may uncover trends and patterns that can offer insights into past performance that are valuable to the committee’s oversight.

For example, with the appropriate technology, GenAI can summarize key issues from the written MD&A or other qualitative disclosures that have been historically addressed so that the committee can see if the issues were successfully resolved or are still ongoing.

Look forward based on the competitive landscape

Audit committee members may also consider using GenAI to evaluate competitors’ current or historical financial statements or earnings call transcripts to create valuable market intelligence. This extraction from competitor documents can be used for performance benchmarking to further enhance the committee’s financial reporting or risk management oversight.

This level of research likely would not have been undertaken by directors before GenAI because of the time and cost it could require.

Getting started

Directors would want to use a private instance of a GenAI platform (e.g., through Azure OpenAI Service, Gemini via Google Cloud). GenAI tools provided through a secure platform, such as Azure or Gemini, are designed to safeguard confidential information and enable uploading larger documents into the generative pre-trained transformer.

Most organizations with private instances of GenAI tools configure them to limit the GenAI from “learning” or integrating the companies’ prompts or uploaded files into its own learning algorithms to help secure the usage and data. With effective prompts, GenAI can often provide citations and pull out trends to help validate that there aren’t hallucinations in the outputs.

Once summaries and key points are generated, review them and keep what is most important. Then, consider developing an additional prompt with the persona of an audit committee member; consolidate and share all the summaries back to the GenAI to formulate strategic questions to consider asking management at the next audit committee meeting. Be aware that with the way GenAI predictive models work, you could submit the same requests multiple times and not get the same answer.