AI Leadership for Boards – The Future of Corporate Governance 1/3
F. Torre, R. Teigland, L. Engstam
McKinsey: in 2017 only 16% of Directors fully understood the impact of technological advances on their organisations and the industry. This means that there needs to be at least some basic level of technological understanding, and if needed – Directors need to be given a crash course and/or refresher course. [MK: the best way to think about it is like this: Directors need to be aware of the company’s risks and decide on the response/mitigation plan. As more risks are becoming technology-driven, understanding the nature of the risk, and making the right decision on the response requires at least some technological savviness.]
The environment is increasingly becoming more volatile meaning the need for more monitoring and faster response. Corporate sustainability (G/ESG) can be achieved only if such monitoring is in place.
AI has a potential to create competitive advantage when employed for strategic and operational decision making. Most of the current AI implementations are operational. [MK: other than the competitive advantage, AI modifies the chain of command in the firm: Board —> CEO —> employee now looks like Board —> CEO —> AI operator —> AI model. It’s important because one can’t order AI to perform a task the way the manager would order an employee. It’s a new kind of management in a way.]
Corporate boards need to develop two competence areas:
Guiding AI operational capability
Supervising AI governance capability.
Ungoverned incompetence – the Board tries to make a decision that turns wrong due to the lack of competence in the Board [MK: as opposed to using all the available information and following a fair process]. It’s a strategic risk for the company unless it’s shifted to the empowered management who can make decisions at the expense of the alignment of interests of the management and the shareholders.
In 2000’s such Board incompetence in digitalisation had contributed to the demise of many Fortune 500 companies, which were innovated away by their more digitally-savvy competitors and smaller firms.
Digital leaders have two relatively high capabilities.
Digital business capability – the use of digital technologies for:
Collaboration across boundaries
Creating new business models and/or customer proposition
Changing the configuration of the firm.
Digital leadership capability – the competencies Boards needs to possess to:
Participate in the identification of digital opportunities
Monitor the risks related to digital transformation
Use of social media and other digital technologies to share knowledge and create a two-way communication with stakeholders.
Still, digital technologies are easier to use for functional tasks, with strategy (i.e., using digital technologies for sustainable competitive advantage and value creation) slowly catching up. A nice proxy metric is a % of existing revenue firms can sacrifice to be replace with new revenue from digital initiatives. [MK: I wonder if it’s possible to reliably determine this % in near time instead of looking back and massaging numbers to arrive at a good consistent story.]
Activities come hand in hand with risks, and Boards are lagging in their understanding of the risks associated with their new shiny digital initiatives, with cyber security risks being the most obvious example.
It’s much easier to understand and address operational digital risks rather than to develop and implement a strategic transformation program. Ideally, the job of the Boards is the latter, not the former (once an appropriate risk management framework is in place).
3/ Guiding AI Operational Capability
Digitisation is both the process of identifying opportunities before they arise and at the same timer responding to these opportunities. This requires sufficient Board competence. [MK: and here’s where I start having troubles with the concept that Boards must drive digitisation. Maybe I’ve been in the internet business for too long – 20+ years – but I find it hard to believe that Directors would be able to tell the CEO which technological areas the company should go into next. I also don’t buy the “experience” argument because by the time a person gets a Board seat, their knowledge and experience will already be outdated: the general will be preparing for the war that has just ended.]
Guiding the Gathering, Harvesting and Analysis of Data
Different nodes in the system are connected and their interaction creates a data trail potentially allowing detecting patterns and understanding behaviours. [MK: let’s not forget that some models can lead to increased inequality, too.]
Datasets represent competitive advantage. Datasets come from: external sources (publicly available), operations (from customers, suppliers, and other partners) and internal sources (the company itself).
MK: The book makes a reference to blockchains, but at least at the time of this writing I think of this “technology” as a tool waiting for a problem, so my exposure to this topic is non-existent. Would love to be convinced otherwise in due time.
Alternative data is anything but the traditional data sets and can be used to enrich them in order to identify new correlations and insights. It comes from non-financial and non-traditional sources: individuals (social media, news, reviews, web activities), business processes (transactional data, compliance, interactions with stakeholders) and sensors (geolocation, activity tracking).
Cleaning, preparing and labelling (aka “tagging”) the data is very time consuming and expensive: it’s not enough to have some data, it must be consistent, accessible and sufficient, otherwise the AI model relying on it will be heavily biased.
Thus, data management requires a quick loop of data origination —> storage —> structure and analysis —> creating insights —> back to collecting more data.
Boards should get the insights in a timely manner to get a feeling of how the business is really going. At the same time, Boards must be aware that data ownership comes with rules and privacy concerns of multiple stakeholders.
Guiding AI-Driven Innovation Strategy
Below are the common patterns for AI use to create value to firms. Such initiatives can and should be implemented in parallel as they have different payback periods. At the same time, implementing AI just for the sake of having AI is wasteful and should be avoided.
Leaner, faster operations – improving efficiency, decreasing costs and better asset utilization (automation, pattern detection, addressing inefficiencies and waste).
Tailored services, products, and advice – customization becomes much cheaper, so customized products can be offered for the same or marginally higher price.
Better self-service with more scenarios – in addition to savings on customer service, many tasks can be automated and often even the customer flow can depend on who the customer it, what their purchasing history looks like and what they are expected to be needing at the time of the contact.
Smarter decision making – the hope that more data will uncover more insights that later on can be used to improve the product or operations.
New value propositions – creating new revenue opportunities through new products and services either augmented with AI, or if the need for them has become obvious as a result of deep data analysis. It’s the “winner takes all” paradigm when the first mover has the advantage of collecting more data, enjoying network effects and increasing market share.
MK: To me this all sounds very shallow. Clearly, it’s not the Board’s job to look at the data and the conclusions to come up with new products and services or improve operations. At the same time, if the AI implementation is brought to the Board as part of the capital spending initiatives requiring approval – then it makes perfect sense for Directors to understand what the managers are talking about.
Implementing of a few AI initiatives shouldn’t be confused with full-scale AI transformation. At the same time, in 2021 successful innovation management almost certainly includes successful AI implementation.
The key factors behind successful innovation culture are:
Engagement of cross-functional teams into the development, implementation, and oversight of the digital strategy.
Establishing a feedback loop (in terms of learnings and the data to train the model). An effect of the loop may be that the model may have to be completely rebuilt if it doesn’t deliver the promised value: the ability to do so is an indication of the presence of the innovation culture.
Top management buy-in into the strategy and the demand from non-technology teams for AI-driven process and product improvements. Communicating the importance of AI and digital throughout the firm and making hiring decisions based on this value.
MK: if you think this sounds fluffy – it probably is.
AI doesn’t come cheap: it requires huge computational power and energy consumption (think environmental sustainability).
Boards need to be talking more about the potential of AI implementation and less – on control issues. This requires building more expertise in innovation, technology and sustainability.
Guiding the Participation and Growth on a Business Ecosystem
AI creates most benefits not for tech firm, but for those building platforms and ecosystems regardless of the industry. More data captured means more value for AI. [MK: I can’t NOT mention the fact that more data can lead to even more institutional biases. On the other hand, it is true that firms can get the biggest bang for their buck once they start doing something new, not when they’ve already perfected it.]
Going it alone is very challenging with AI: building the initial capability is often better to do with partners; hiring own AI staff is challenging due to the global shortage of expertise in this growing area.
The challenge for Boards is understanding complex adaptive systems: understanding of parts is not sufficient for understanding their interactions and the business value created in the process. Understanding the benefits of the ecosystem’s components to its customers as well as charting the evolution of the ecosystem is a very time-consuming recurring activity for the Boards.
The speed of development and evolution of ecosystems can’t be captured by regular Board meetings that occur every 2-3 months; it’s helpful to have an Innovation Committee, but this should not stop Boards from exchanging notes more frequently outside the meeting cycles.