This article first appeared as part of the Boardroom Sensemaker newsletter – The Observer’s weekly newsletter for the board chairs, directors and executives in charge of Britain’s biggest businesses. To receive it in your inbox, featuring content exclusive to the newsletter, sign up here.
More than four in five company boards are now debating which of their decisions should stay human-led and which can be handed to artificial intelligence, according to a new survey of over 400 directors, chief executives and finance chiefs.
So what? Only four out of five: what on earth is the other fifth doing? AI is reshaping every corner of business, and there is no obvious reason the boardroom should be the one room it cannot enter.
In the same survey, by boardroom advisory and software firm Board Intelligence:
•
86% confess their own rigid or inconsistent decision-making produced a delayed, rushed or poor call in the past six months; and
•
just 37% regard their board as essential to creating value.
This is an institution that needs help. The practical question is whether AI will be able to provide it.
Redirection. Survey respondents were split on whether, or how, it can: while 49% of boards have moved from discussing AI’s role to implementing it, 40% expect only marginal change to their own processes over the next five years, whilst 8% think boards should be “completely reimagined”.
Completely? Perhaps directors will be freed to enjoy longer lunches, letting the meeting be conducted by their digital twins? Already, Sam Sidhu, CEO of America’s Customers Bancorp, has deployed an AI clone to deliver an earnings call (admittedly a marketing stunt), and the bosses of Zoom and Klarna have had avatars front investor updates. Will activist shareholders one day propose slates of AI bots to replace underperforming human directors?
Where the buck stops. AI-driven efficiency gains through better analysis of board documents and financials and sharper agendas may prove valuable, but the fundamental job of a board is to be where the buck stops: the body that can be sued, sacked or shamed when its judgment fails.
That buck cannot be handed over to AI, because accountability must rest with humans capable of suffering the consequences. Yet if AI could help improve the quality of boards’ judgments, that might truly transform their performance.
The chummiest room. The deepest boardroom failing has always been social rather than technical. Many studies have found board consensus is often driven by the preservation of collegiality rather than rigorous evaluation – with self-censorship, peer conformity and deference to dominant chief executives.
Newsletters
Choose the newsletters you want to receive
View more
For information about how The Observer protects your data, read our Privacy Policy
Attempts to change these dynamics by, say, drafting in outsiders or splitting the roles of chair and chief executive have too often been frustrated by the social pressure to not rock the boat.
Advocatus diaboli. Here an AI might genuinely earn its place in the boardroom. The idea of a purpose-built AI “devil’s advocate”, designed to surface counterarguments and flag signs of groupthink, is already being urged on boards by business schools and explored in a fast-growing academic literature.
The AI’s edge would be temperament as much as brainpower: a bot has no career to protect, no reappointment to angle for and no weekly round of golf with the boss to preserve, making it blind to the pressures that gag boardroom dissent.
Perhaps the boardroom bot could even be invited to cast an indicative vote (recorded and reasoned, but not counted), forcing any elephants in the room to be addressed without surrendering real authority to a machine.
The new conformity. This idea is still at the concept stage and would not be without risk. Some scholars warn of potential “epistemic capture”, in which boards cede their knowledge authority to an AI seen as an oracle of objective truth, breeding a fresh “algorithmic groupthink” as directors nod along to the machine rather than the chief executive.
A different danger, as PwC’s Paul DeNicola has noted of human sceptics, is that the AI becomes the “chronic dissenter” the room learns to ignore.
What’s more… When BP abruptly removed its chair, Albert Manifold, in May after less than a year, citing unspecified concerns about “governance standards, oversight and conduct”, the remedy was not an algorithm but a senior independent director and a board willing to wield their ultimate sanction. Which points to the oldest question in governance: who guards the guardians?
A bot built to challenge the chief executive must still be commissioned, tuned and, when it grows tiresome, switched off by somebody. If that somebody is management, rather than, say, shareholders, dissent might be easily programmed to lean in the boss’s favour.
The original advocatus diaboli was a Vatican official appointed to argue against candidates for sainthood, a role that could be effective because it was the Church, not the would-be saint, who appointed him.
Photograph by Fairfax Media via Getty Images