Several times in the past six weeks, I have been asked to share some thoughts on artificial intelligence and board work; specifically, the impact of emerging AI capabilities on corporate governance and, even, the need for board of directors. The rapid emergence and now widespread awareness of ChatGPT has been a catalyst for many of these enquiries, it seems. I have been fascinated by the unfolding situation, not only because of a longstanding interest (I studied artificial intelligence at university nearly four decades ago), but also the speed by which awareness has spread, and expectations climbed to such stratospheric heights, is unprecedented. Claims have been made that computer-based tools will soon supplant the need for human directors and, with it, board meetings. Some, especially those with jaundiced perceptions of boards, their work and any value they add, have confided this may be a good thing. Others have reserved judgement—for now at least—saying the situation is far too fluid and complex to make anything approaching an informed or reliable decision, much less widespread change.
That so many people are questioning 'conventional' corporate governance practices feels a little bit like ground hog day. While I do not claim any particular expertise in the topic of artificial intelligence, I have read widely, asked many questions (of myself and others) and pondered both the purported capability and potential impact (of artificial intelligence) on board work.
The departure points for my enquiry has been, as always, definitional. What is artificial intelligence, and what conception of governance does one hold? My responses to these questions are as follows:
- Artificial intelligence: the development of computer-based capability to perform tasks that normally require human intelligence. In effect, the simulation of [aspects of] human intelligence by machines.
- Governance: the provision of steering and guiding an organisation [towards an agreed goal], by the board of directors. In effect, governance is the work of the board—the means by which companies are directed and controlled.
So, the proposition to be considered is, "Can a computer replace a social group charged with steering and guidance an organisation in a complex and dynamic environment?"
Those people wondering whether AI might be a viable mechanism to support or even replace boards have much to ponder. What is the role of a board of directors in companies? How might the operating context beyond the organisation be assessed? Where does accountability for statutory compliance and overall performance lie? And, to whom should the Chief Executive and management of the company report? If one holds the view that the board is the ultimate decision-making authority within a company (a responsibility delegated by shareholders), and that this (decision-making despite uncertainty and ambiguity) is 'core business', the board has a vital role to play.
My early training in computers and technology taught me that computers respond to instruction; they cannot 'think' autonomously or handle ambiguity, and they lack feelings and intuition. They do what they are 'told'; if the 'telling' is poor, the result is likely to be poor: the phrase "garbage in, garbage out" springs to mind.
But that was then. Computing power is far greater today than it was even five years ago, much less forty. Has the evolutionary development of computing capability reached the point whereby computers can displace humans? For a large and growing list of tasks and activities, yes, of course. The analysis of data is a relevant case in point. But for many other enquiries, the answers remains a resounding no. How might a computer make sense of the unspoken feelings, intuition and biases of staff, customers and board directors, and reach a credible decision? For this, a much higher order of capability is necessary. And, with that, I stand with those reserving judgement.
What of the future? AI may become a viable mechanism to expedite board decision-making, of course. But the likelihood of directors being supplanted any time soon is low (those failing in their duties excepted). For that, artificial general intelligence (AGI) is likely to be necessary, and some moral and ethical questions will need to be resolved as well. If that is achieved, I may take a stronger position.
Regardless of whether this muse is sound or not, directors, shareholders, regulators and their various advisors need to be alert, because the situation may change quite quickly.
