Author: William Vorhies
Summary: Communicating with your Board of Directors about AI/ML is different from conversations with top operating executive. It’s increasingly likely your Board will want to know more and planning that communication in advance will make your presentation more successful.
Whether this comes in the form of being summoned to discuss and explain, or whether you, along with your CEO decide to proactively communicate, it will pay to consider in advance what that communication should include and what the message should be.
This will not be the same as addressing senior operating management and it’s important to start with understanding what your Board of Directors does and doesn’t do.
The Job of the Board
From an objective standpoint, for-profit Boards have a fiduciary responsibility to shareholders that requires two major types of activity, one advisory and the other oversight. In brief, that would typically include the following:
- Hire and compensate the CEO.
- Approve the corporate strategy.
- Test the business model. Monitor the company’s products, services, and programs.
- Approve the budget and major asset purchases.
- Ensure adequate resources.
- Ensure the integrity of published financial statements.
- Protect the company’s assets and reputation.
- Ensure the company complies with laws and regulations.
- Represent the shareholders and ensure actions of the company are designed to protect and drive shareholder value.
The most important issue here is that the Board does not formulate the strategy or manage the company. Neither do they direct what technologies or business models should be adopted.
Their role is fundamentally a conservative one, to review the work of others and protect the interests of the shareholders.
On a subjective basis, while Board members may be very senior in their experience of your and other industries, they may not be as well informed as you and senior management on new technologies and how to exploit them.
Second, their meetings are relatively short and infrequent, so unless specifically asked it’s unlikely you’ll get more than two hours to make your case, and even that may be generous. The best case may be that there is a standing committee of the Board for technology, but these are relatively uncommon.
In short, how do you keep your message relatively short but sufficiently deep to create understanding. What topics should you cover?
I would argue that this conversation actually requires two separate sessions, first one to educate Board members about what AI/ML is and what it can do, and a second more substantive conversation about whether the company’s approach to utilizing AI/ML is correct.
While the education session will clearly cover many basics that don’t need to be detailed here, it should include these topics:
- AI/ML consists of several distinct technologies, not all of which are as ready for commercial adoption as others.
- ML scoring models and forecasting models are a mature technology.
- IoT and edge computing applications are mature.
- Parts of NLP such as chatbots are rapidly maturing and can be confidently implemented.
- Most image processing or image classification applications remain bleeding edge and their implementation risks remain substantial.
- Really cutting edge technologies like GANS or reinforcement learning are simply not ready for prime time.
- Despite the popular press, AI/ML is not yet the new electricity. It is another in a long string of technological advancements that need to be embraced. However, just as with all previous technologies, it is the way the organization deals with its implementation that determines what value will be realized.
- Any implementation of AI/ML is dependent on the existence of suitable data. Acquiring and maintaining that data has implications for cost and for the IT organization that need to be figured into all plans.
Organizing the Strategic Conversation
In this conversation I suggest you think like a Board member and divide this into two broad topics along the lines they might think of it:
- How do we make sure we get value from our AI/ML investments?
- What risks do we need to be aware of and how can they be mitigated?
The Value Conversation
The first thing to understand, is that unless your company is one of the very few true leaders in adoption, neither your Board or your management will be thinking outside of the current business model. The question on their minds will be how to use these tools to improve current operations. If you need a refresher look back at our conversation that differentiates ‘AI Inside’ or ‘Applied AI’. That list of opportunities should include at least these:
- Operational efficiency (including IoT opportunities and robotic process automation).
- Business expansion. Discovering new offerings or better channels through the use of more data and better analytics.
- Enhanced customer experience.
- Better more efficient marketing.
- Better pricing.
Just as important as Applied AI is the potential for evolving wholly new business models around data services and platform strategies. This will be a dramatic diversion from typical thinking about your company’s business model and its ability to hold off future competitors.
This is as much a part of the ‘risk’ conversation as it is a part of the ‘value’ conversation. If your competitors beat you to a platform strategy you might find your company just another commodity supplier in someone else’s AI/ML platform ecosystem.
The Risk Conversation
To repeat, AI/ML leaves your company open to be outflanked by competitors, especially in the area of platform strategies.
Separate from this however are two relatively operational risks.
- There is significant operation risk from piecemeal implementation. It would be extremely common for your company to already be using AI/ML in many pockets throughout the organization.
Like CRMs and ERPs before this, the failure to create a central standardized approach and strategy will almost surely lead to greater pain later.
As those individual applications of AI/ML become increasingly important to the operation of their business units, they are likely to have been based on different analytical platforms, programming languages, and data sources and feeds. When the time comes that the case for standardization becomes overwhelming, the cost of tearing these out and starting over will be extremely high.
Best to start now with agreement on an overall AI/ML strategy, standardized platforms, and methods of data governance. This also supports best use of financial and human assets.
- Second is a question already frequently asked at the Board level. What is the risk of a data breach or unintended inclusion of bias into an important operating model? Both of these are actually pretty likely and the importance of the risk will depend on your industry. If you are in a heavily regulated industry like finance or healthcare, or if you hold a dominant position of trust with a large number of individual users (social media) it will be best to start thinking about your response now.
About the author: Bill is Contributing Editor for Data Science Central. Bill is also President & Chief Data Scientist at Data-Magnum and has practiced as a data scientist since 2001. His articles have been read more than 1.5 million times.
He can be reached at: