---
title: AI and the CEO: Choosing the Bets That Matter
date: 2026-06-21
description: The building can move to a machine; the choosing and the answering cannot. Where AI changes the CEO's role, and where it changes almost nothing.
author: map[email:mario@mariothomas.com name:Mario Thomas]
canonical: https://mariothomas.com/blog/ai-board-director-ceo/
---

The market now treats visible AI adoption as proof a company is driving forward through innovation, and the chief executive is the one expected to show it. Being seen to adopt, not misleading the market, and choosing well are three demands held at once. In this article, I argue that AI does not rewrite the chief executive's duties; it changes the conditions under which they are discharged. The task is to choose the few bets that matter, change how the company works around them, and answer for them without delegating the accountability.
<!--more-->
{{< image3 src="ai-board-director-ceo" type="photo" alt="A corner executive office in dark mahogany panelling at night, lit warm on the left by a stained-glass Tiffany lamp beside a porcelain teapot and cup, where a spread of closed leather dossiers fans across a polished desk and one is squared up in the lamplight before an empty high-backed leather chair with a fountain pen resting on it as the chosen bet, the rest angled away into shadow, and cool blue light falls from floor-to-ceiling windows onto a night city skyline and waterfront beyond, a visual reframe of the few bets a chief executive draws from many and answers for, where a machine can lay out the options but only a human can choose (Image generated by ChatGPT 5.5)" width="735" height="413">}}

{{< audio2 src="mp3/ai-board-director-ceo.mp3" >}}

According to IBM's 2026 [CEO Study](https://www.ibm.com/thought-leadership/institute-business-value/en-us/c-suite-study/ceo), **69%** of chief executives say AI is already changing what they consider core to their business. The market reads that adoption as proof a company is driving forward through innovation. A [FactSet](https://insight.factset.com/more-than-65-of-sp-500-earnings-calls-for-q4-cited-ai) review found the term "AI" cited on a record **331** S&P 500 earnings calls in a single quarter, more than two-thirds of the companies that reported. The returns do not match the noise. PwC's 29th [Global CEO Survey](https://www.pwc.com/gx/en/issues/c-suite-insights/ceo-survey.html) found that **56%** of chief executives report neither revenue gains nor cost reductions from AI, and only **12%** report both, while its 2026 [AI Performance Study](https://www.pwc.com/gx/en/news-room/press-releases/2026/pwc-2026-ai-performance-study.html) found that **74%** of AI's economic value is captured by just **20%** of organisations. McKinsey reads the work as [80% business transformation and 20% tech transformation](https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/how-the-best-ceos-are-meeting-the-ai-moment), with the distance between the leaders and the rest widening, which makes where a chief executive chooses to commit the decisive question. The CEO sits on both sides of this at once, the executive director who deploys the AI and the public face answerable to the market for it. The conditions have shifted under every one of those duties. What has not shifted is who answers for them.

## The principle the CEO operates

The constitutional foundation is settled, and AI has not amended a line of it. The FRC's 2024 [UK Corporate Governance Code](https://www.frc.org.uk/library/standards-codes-policy/corporate-governance/uk-corporate-governance-code/) names the chief executive as the executive leadership of the company's business, distinct from the chair who leads the Board. The Code requires a clear division of responsibilities between the leadership of the Board and the executive leadership of the business, holds that the two roles should not be exercised by the same individual, and requires that those responsibilities be set out in writing and made public. [Cadbury](https://cadbury.cjbs.archios.info/report) framed the chief executive in 1992 as the head of executive management the Board monitors. The [Companies Act 2006](https://www.legislation.gov.uk/ukpga/2006/46) binds the chief executive as an executive director through the general duties, including the duty to promote the success of the company and the duty to exercise reasonable care, skill and diligence.

The chief executive sits on both sides at once, the executive director accountable to the Board for the organisation's AI and the public face answerable to the market for it. That duality sits at the centre of the role, and within it agency and accountability behave differently. An AI model can run the analysis, an agent can execute a workflow, and a chief technology or data officer can own delivery, and none of those is illegitimate. What cannot move is accountability for whether the bets paid off and whether what was said about them was true.

So the texts have not changed, and not one of the existing duties is relieved. The chief executive's responsibility to adapt the organisation to its environment was always part of the remit. What has changed is that the environment now demands a complete reengineering of the business rather than an adjustment to it. The chief executive therefore carries every existing accountability and, at the same time, leads a remaking of the work itself.

## The strategic duty: choosing the bets that matter

Under conditions of hype, the defining act of the role is the choice of where the organisation commits, and the evidence says that choice is where value is won or lost. The PwC numbers are not a verdict on the technology but on selection: when **74%** of the value sits with **20%** of firms while most chief executives report no gain at all, what separates them is not how much they spent but what they chose to build and how few things they chose to build well. McKinsey's reading of [where AI creates value](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/where-ai-will-create-value-and-where-it-wont) points the same way. Productivity gains spread thinly across an industry tend to be competed away, while durable advantage comes from reshaping how the business works, and the early movers who reshape first compound their lead. Its 2026 [technology agenda](https://www.mckinsey.com/capabilities/mckinsey-technology/our-insights/mckinsey-global-tech-agenda-2026) lands in the same place: the pattern is not that the leaders spend more, but that they spend with intent. **AI advantage is a function of choice, not of spend.**

The act is the chief executive's, and worth keeping distinct from the one beside it. The chief executive chooses which bets the organisation makes and stakes its direction on them. The CFO stewards the capital and answers for whether the bets earned their keep, a responsibility I set out in [AI and the CFO](/blog/ai-board-director-cfo/), while the Board approves the allocation and holds both to account. Conflating the choice with the stewardship blurs where accountability for each one sits.

The depth of the choice is what makes it hard. The bets that matter are bets on remaking how the work is done, not on tools bolted onto it, which is the argument of [The Great Remaking](/blog/the-great-remaking/). Choosing well means choosing where to remake the work rather than where to digitise it, a harder judgement than approving a procurement.

The seductive foil here is the founder-chief executive of an AI-native or frontier company, whose advantage is conviction, personal fluency and concentrated bets made at speed. It is the wrong model to imitate. The chief executive of an established company cannot become that founder and does not need to. What the evidence supports is available to any of them: the fluency to judge a bet, the discipline to choose a few, and the accountability to stand behind the choice.

Fluency is part of that task, a duty rather than a nicety. A chief executive who does not use these tools cannot judge what they are worth, and the people most fluent with them are often the most junior in the building. This is decision fluency, not coding. Delivery can be delegated to a technology or data officer or to the CFO, but under the Code the Board still holds the chief executive to account for whether the bets paid off. And because choosing well is what makes honest signalling possible, the choice carries the market duty with it: strong bets afford modest claims, weak ones force stronger ones.

## The operational duty: changing how the company works

A chief executive does not get credit for buying AI. The credit, and the value, come from changing how the company works. The returns sit with the organisations that redesign the work rather than layer tools onto unchanged workflows, and IBM's 2026 study puts numbers on it: **77%** of chief executives say the boundary between business and technology has become obsolete, and those redesigning how cross-functional teams work are more than **twice** as likely to deliver on their objectives. The decision that matters most has stopped being which tool to license and become how the work and the organisation around it are arranged.

That is an operating-model question before it is a technology one. It means rethinking who owns which decisions, and governing the spread of agents across the business so that proliferation does not become something no one can see or account for. [Minimum Lovable Governance](/toolkit/minimum-lovable-governance/) is the operating principle: light enough to keep the speed the choice was meant to buy, structured enough that the organisation can attest to what it has built. The aim is not to slow the spread but to keep it legible.

Read from the chief executive's chair, the **56%** who report no gain is not a statement about the technology. It is a statement about operational follow-through. The model was capable; the work around it did not change enough to convert the capability into value. This is also where the advantage compounds. The organisation that can test, learn and scale faster than its competitors widens the distance over time, which is why the operational work is not a tidy-up after the strategic choice but the place the choice either pays off or does not.

## The market duty: signalling without overclaiming

This is the duty that is uniquely the chief executive's, and the one where the law bites hardest. As the principal public voice and the sponsor of what the company tells the market, the chief executive sits inside the duty not to mislead. The UK footing is established and live. The FRC Code requires the annual report to be fair, balanced and understandable, and the directors' responsibility statement makes that a personal undertaking. The same Companies Act duties to promote the success of the company and to exercise reasonable care apply to the claims as much as to the strategy. For firms within the financial regulator's perimeter, the FCA requires communications to be [fair, clear and not misleading](https://handbook.fca.org.uk/handbook/COBS/4/2.html), with personal accountability under the [Senior Managers and Certification Regime](https://www.fca.org.uk/firms/senior-managers-certification-regime). And the CMA's powers under the [Digital Markets, Competition and Consumers Act 2024](https://www.legislation.gov.uk/ukpga/2024/13) reach misleading claims of every kind, AI included, with penalties of up to **10% of global turnover** for the business and up to **£300,000** for individuals.

The pressure to overclaim is real: AI has become shorthand for value, and the record number of earnings calls citing it is the visible form of that pressure. The hazard is that overclaiming has a name, AI washing, and it is being pursued under existing law rather than any new AI-specific rule. The clearest leading indicator is the run of US enforcement: the SEC's [first actions](https://www.sec.gov/newsroom/press-releases/2024-36) against two investment advisers in 2024, its first against a public company in 2025, and an AI-app founder charged personally by the SEC and the Department of Justice that same year. These are not the UK rule, but they establish that disclosure and anti-misrepresentation law already applies to AI claims, and that individuals are being named. UK exposure runs differently. Without the United States' developed securities class-action regime, the pressure falls on the regulators and, decisive for this role, on the individual senior manager, where the chief executive is personally exposed. **Being seen to adopt AI is an exposure, not a strategy.**

The resolution is the one planted in the strategic duty, which keeps this duty subordinate to it. The discipline of choosing well is also the defence against overclaiming. A chief executive with strong bets can describe them plainly and let them speak; one with weak bets is tempted to dress them up, which is the conduct the law is now reaching. Honest signalling is a consequence of having chosen well, not a separate skill.

## The part that does not move

Some of the role of the chief executive AI barely touches, and it is the part the anxiety most often assumes is dissolving. The work that remains stubbornly human is the leadership work. As McKinsey puts it in its study of [leadership in the AI era](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/building-leaders-in-the-age-of-ai), generative AI cannot set aspirations, make the hard calls, build trust among stakeholders, hold people to account, or originate truly new ideas, and the leaders who thrive are the ones who use AI to "think with them, not for them."

Naming this plainly matters, because the core of the role is not in question. The framing of an audacious brief, the holding of the line when the early iterations are messy and the temptation is to abandon the bet, the trust with the Board and the market that lets a chief executive ask for patience and be granted it: these are the parts AI reaches last, if it reaches them at all. A model can inform them at the edges. It cannot form the aspiration, and it cannot bear the responsibility for the call.

## The line that does not move

This is the principle the series has carried from the start, applied now to the role that sits on both sides of it. [AI and the Director](/blog/director-ai-governance-playbook/), [AI and the Chair](/blog/ai-board-director-chair/) and [AI and the Company Secretary](/blog/ai-board-director-secretary/) traced it through the oversight roles, whose duty is to the Board's collective accountability. [AI and the CFO](/blog/ai-board-director-cfo/) was the first executive officer, accountable for the integrity of the numbers and the stewardship of capital. The chief executive completes the set as the one office answerable on both counts and to the market besides.

The chief executive can delegate the building, the delivery, and even the drafting of the message. What cannot be delegated is the answer: for whether the existing obligations were kept, for whether the business was remade to meet its environment, and for the truth of what was claimed about it. AI can contribute to the work of the reengineering and the work of the report. It cannot be accountable for whether the business met its moment, or for whether what was said about it was true. AI changes who builds and who signals. It does not change who answers. That is the line that does not move.

{{< campaign "ai-board-director-ceo" "hello@mariothomas.com" "Hello" "Let's Continue the Conversation" "Thank you for reading about the chief executive's duties under AI, and the choice that sits at the centre of them. I'd welcome hearing how this looks from where you sit, whether you're working out which few bets to commit to rather than spreading spend across many, redesigning how the work itself is done rather than bolting tools onto it, or weighing how to signal adoption to the market without claiming more than the evidence supports. I'd value understanding where you're finding that the accountability cannot move, even as the doing of the work does." "Thank you for submitting your details, here's what you provided:" "Click send to share your input with me." >}}