AI and the Board’s Responsibility to Challenge it: The Buck Stops Here

The rise of AI obliges board members to reinforce their commitment to critical thinking.

Since at least 2023, Geoffrey Hinton, Nobel laureate, sometimes called “the godfather of AI”, has been speaking out about its dangers. Speaking of the big technology businesses, earlier this year he told Fortune magazine, “For the owners of the companies, what’s driving the research is short-term profits”. Developers, he said, are more interested in solving immediate problems than in AI’s long-term consequences.

Those developers aren’t alone. Many people, from job applicants writing CVs all the way up to global enterprises, are using AI as a quick fix. And if that’s the case, it’s incumbent on boards to take the long view as well as the short one.

Practical AI challenges and goals

It’s safe to assume that few, if any, companies these days are greenfield sites for AI. Even if it isn’t yet being applied to the big levers of power, it’s being used by employees every day and in all kinds of ways.

Which means that the first challenges the board must address are practical ones. In governance, for instance: how do organisations achieve oversight on the use of AI company-wide? Do they develop their own guardrails, or should they be led by regulation or industry best practice? Who will manage the human use of bots – and who will oversee bots’ use of other bots?

For this article we spoke to a current NED with extensive global executive and non-executive experience, who said:

Even more fundamentally and at a macro level, boards need to ask: what are our goals for the application of AI? Because let’s face it, even though companies often argue that they welcome AI for its ability to transform long-term strategy, the temptation to put it to work for short-term tactical advantage, as Geoffrey Hinton said, is very strong. Even Big Tech, which invented it, is using it to achieve significant headcount reductions and productivity gains. Over 100,000 tech workers were laid off in 2025 alone, with AI cited as a primary driver in more than half the cases.

Business dilemmas

The issues AI creates can also be practical, real-world business dilemmas and these can be harder for boards to resolve. For instance:

The danger of homogeneity

You’d think that the larger the data set and the more comprehensive its frame of reference, the greater would be AI’s creativity and the broader the range of its recommendations. There is evidence to suggest otherwise. A 2024 MIT paper described, first, the Platonic Representation Hypothesis, which argues that “neural networks, trained with different objectives on different data and modalities, are converging to a shared statistical model of reality in their representation spaces”; and second, the Simplicity Bias Hypothesis, which maintains that “deep networks are biased toward finding simple fits to the data, and the bigger the model, the stronger the bias. Therefore, as models get bigger, we should expect convergence to a smaller solution space”.

In short, the more that deep neural networks develop, and the more businesses depend upon them, the more important it will become to frame propositions fully and fairly. Get it wrong, and the danger grows that proposed solutions will seem specific, but may in fact be generic.

The danger of overconfidence

The dilemma for boards thinking of committing to AI is whether they – and every other part of their businesses – will act on its recommendations by virtue of that commitment, and of the investment the organisation has made in it. If it’s been right many times before, for this organisation or for others, strategically or operationally, will it be safe to assume it’s right this time too? The answer, obviously, is no: now more than ever, board directors need to maintain their capacity for critical thinking – and that means ensuring that existing oversight frameworks also accommodate AI-based decision-making.

The optics

Even though boards may be able to justify using AI for short-term tactical advantage to their shareholders, there is no guarantee that the opinions of any stakeholder group are immutable. Even shareholders may revise their views of a leaner, higher-margin operating base if they see perceive potentially negative financial consequences.

When making decisions based on AI, boards need to be mindful of how their actions will be received by stakeholders – and they should be prepared to respond if need be.

The danger of simultaneously streamlining supply and crippling demand

Using AI to slash the workforce and improve margins isn’t just a simple short-term business gain. In fact, it may not be a short-term gain at all – and what’s more, it may also potentially be a long-term business problem.

An article in Forbes magazine earlier this year quoted a prediction that research organisation Gartner made in February: “… by 2027, half of companies that attributed headcount reductions to AI will rehire staff to perform similar functions, often under different job titles.”

Which begs the question: why? Gartner’s Kathy Ross, a Senior Director Analyst, said: “Most recent workforce reductions were influenced by broader economic conditions rather than automation alone. As organisations encounter the limits of AI and rising customer expectations, they will need to reinvest in human talent to sustain service quality and growth.”

Firing and rehiring are not only inefficient but bad for goodwill – and in the meantime, by using AI as an argument for layoffs and perceived short-term gains, companies risk collectively eroding the spending power on which they depend for their income.

It’s not just about generic spending power either. It’s also about repercussions specifically for the brand: a business that displaces workers may be less attractive to customers, particularly if those customers are ex-employees, temporary or otherwise. And of course, the more unemployment rises, the more government income tax revenue decreases, worsening the socio-economic climate in which the business operates.

Using AI to reduce the workforce permanently also potentially creates a future talent gap, because smart technologies often replace middle-ranking white-collar functions – and people in these roles represent a big part of the pool from which future executive team leaders and even board members are drawn.

The buck stops here

If there is one point that is implicit in all the challenges and dilemmas we’ve highlighted here, it’s this: the importance of objective and impartial analysis, and of keeping oneself informed when AI is moving so fast. The same NED told us: “It is the responsibility of individuals to keep themselves up to date. Board members shouldn’t be waiting for the board to teach them. Engaged and interested directors are keeping up to date to ensure they ask the right questions of management.”

AI could well be exhilarating, and we could well be entering golden years. Some people are even asking if bots have a role on the board. But that doesn’t mean that anyone gets a free pass to relinquish responsibility – board members least of all. They need to stay abreast of AI developments in theory, in practice, and specifically within their own organisations.

Even more importantly, they need to continue to apply the critical thinking that probably earned them their roles in the first place.