28/07/2023
No board pack or agenda would include the simplistic question “Should we buy some Artificial Intelligence?”. If you have one in front of you, then it may be best to quickly consider an adjustment to your portfolio as a director. If, however, you and your fellow directors are thinking about and discussing how AI in its various forms might affect your organization - then you are most likely on the right track.
Artificial Intelligence is a term used to describe technology that can replicate or enhance the way human intelligence approaches tasks and that is sometimes able to constantly improve that approach without intervention (learning) – mirroring some human traits. A common understanding of AI varies from science fiction tropes (like Skynet) right through to a good grasp of how Google Maps can predict travel times or how Netflix recommends your next binge-watch.
FINDING OPPORTUNITIES IN A FAST MOVING LANDSCAPE
With advances in computer hardware processing capacity (things like scalable cloud infrastructure and graphics processors), the challenge of building and testing AI models is fast becoming less expensive and more accessible. This in turn means that research and development into the various types of AI is moving at a very high speed. Consider Speech Recognition technology for example. We might be quite used to Siri or Alexa in our everyday lives. But this technology has been around for a long time, making gradual improvements until the 2010s when competition, focused R&D and available resources allowed it to close the gap with a human level of accuracy in 2017 in just a few short years. Last year, OpenAI released a model with even better accuracy of ChatGPT - which not only transcribes but can translate into other languages as well. Furthermore, consider the challenges presented right now in the boardrooms of Adobe, Getty Images or Shutterstock. Three companies that supply images and who are facing new technologies like DALLE - able to generate realistic images from a written prompt. There's no doubt that we'll see companies' business fundamentals undergo major upheaval and perhaps ultimately disappear altogether.
EXPECT THE UNEXPECTED: SUDDEN PIVOTS, DISRUPTIONS, AND STOPS
As with anything else moving at a very high speed, there exist opportunities for impressive outcomes but there are also attendant risks of sudden and painful deceleration. For a board, this simply means aspects of the operating environment can change very quickly. This could manifest in many ways; the sudden appearance of a disruptive competitor is just one example. Consider the challenges presenting right now in the boardrooms of Getty Images or Shutterstock. Two US listed companies that supply images and who are facing new technologies like DALLE - able to generate realistic images from a written prompt. There's no doubt that we'll see companies' business fundamentals undergo major upheaval - the best example of which, as everyone knows, was Kodak's retention of film cameras - and perhaps ultimately disappear altogether.
Whilst the probability of a thinking machine helping a disruptor to take big bites out of your market share may be low today, the probability of AI altering the requirements and behaviors of the stakeholders you serve is significantly higher. It already makes sense for a board to regularly consider the organization’s most valuable assets and understand how customer value is best derived from them.
APPLYING A GOVERNANCE LENS
It is only one step further for a board to consider how the organization (or a competitor) might reasonably make use of AI to bring a step change to the value created by the organisation and the way customers are served. Questions like this are one approach for a board to usefully fit the subject into the board agenda.
This is where curiosity matters - an essential director’s quality. Advances in AI as well as practical examples from many different sectors abound and are available to be read and understood – usually in plain language. A little searching and reading can significantly upgrade understanding and the quality of discussions around the table.
All the possible ways in which different types of AI might affect the business model can be a fun and enlightening discussion, which may surface exciting opportunities. Where things get more complicated is where a board wishes to consider investment in AI or adoption of relatively new AI technology into the business. Of course, it is the same as considering any other investment in technology which requires discernment and foresight.
SUCCESSFUL AI PROJECTS HAVE IMPORTANT PRECURSORS
Before spending time and effort on technology choices and constructing a detailed business case for investment a board should consider the wider context of requirements for success – which often includes a need for a bare minimum state of internal readiness before any AI project can begin. For example;
Automating claims processing for insurance is a current real-world application of AI technology. The efficiency comes from reducing the manual interpretation and input required. But readiness will involve having historical data to train the model on, that data being of sufficient quality and also understanding the long-term requirements (human and technical) to maintain the model.
Thus, some good high-level questions for a board agenda item considering an AI opportunity for an organization would be;
- Which data sets will the project need access to?
- If it is a machine learning model, does training data exist yet?
- How will the proposed project integrate to existing workflows and technology?
- Do we have the resources to maintain the model?
- Will AI make the most impact or might some other small change?
It is not uncommon for organizations to investigate the use of AI only to find that they are not ready. This is hardly a fail though, simply a recognition of the potential for the technology and an option to decide whether to close the gap or not. Broadly speaking, the greater the digital maturity of an organization the more likely they are to be ready to adopt different forms of AI.