--- title: "AI & the Workforce" date: 2026-07-12 description: The workforce question is not how many roles AI removes, but whether people build capability that earns premiums or credentials that incur penalties. author: Mario Thomas canonical: https://mariothomas.com/briefings/ai-and-the-workforce/ --- ## Start here Two short essays: the first frames what is actually at stake in the workforce question, the second maps the most useful route through the briefing. ### More Valuable Work, Not Less Work The people side of AI adoption is where the structural evidence is least subtle: talent, skills, redeployment, and what happens to the employment relationship as AI takes on more of the work. PwC's 2025 Global AI Jobs Barometer puts the wage premium for AI skills at **56%**, while Harvard research published in 2025 found that workers targeting high-AI-exposure roles without genuine capability development face a **29%** earnings penalty. The same roles, opposite outcomes. The difference is the quality of the capability investment, not access to the tools. Most of the Boards I meet approach this from the wrong end. The AI business case arrives framed as headcount reduction, the training budget buys tool familiarity and completion certificates, and workforce resistance is assumed rather than tested. The evidence contradicts all three. Workers are readier for AI than the resistance assumption allows, token retraining delays displacement rather than preventing it, and the organisations capturing the largest gains are those that redesigned work around AI rather than dropping tools into unchanged processes. The position these articles take is that AI's primary workforce value is releasing the intellectual capital trapped in undifferentiated work, and that the right measure of success is redeployment, not reduction. That claim carries obligations with it: capability development that builds verification and judgement rather than credentials, expertise pipelines that are augmented rather than replaced, and a transition managed as partnership rather than extraction. Not every skill deserves preservation, but the atrophy has to be chosen, not stumbled into. There is a Boardroom dimension too. The Institute of Directors now positions AI competence as a core NED responsibility, and the divide opening in the workforce has an exact equivalent around the Board table: directors who can evaluate AI strategy independently, and directors who ratify management narratives they cannot assess. Read this briefing and you should be able to make one judgement with confidence: whether your organisation's workforce investment is building the capability that earns premiums or collecting the credentials that incur penalties, and whether your Board could currently tell the difference. ### From the Divide to the Design Take [the articles](#core-reading) in the order given. The core sequence opens with the talent bifurcation, because the premium-and-penalty evidence is the sharpest way into the whole subject: it establishes that the workforce question turns on the quality of capability investment, not the quantity of it. From there, the upskilling article sets out how genuine capability is built, segment by segment, and the redeployment dividend answers the question that follows: once AI frees your people's time, where should that capacity flow? Governing the Redeployment Dividend then shows what happens when that capacity is left unowned: the hours are saved and then reabsorbed, and capturing them is a Board discipline. The sequence then widens the lens. The three-hour work week is the earliest article here and deliberately speculative: it asks what happens to hours and compensation when the value of human work concentrates into judgement, and what Hollywood's AI agreements suggest about sharing the benefits. The future of AI expertise then brings the argument back to organisational design: who builds and manages AI-capable teams, and why the questioning mindset matters as much as technical depth. The further reading extends the argument in several directions: what happens when employees own the most capable AI in the building, how the remaking of the four dimensions of work runs at different speeds, why culture rather than capital determines adoption, where accountability sits when work is delegated to machines, what the oversight labour AI adds costs when no business case has priced it, and why the verification premium means that delegating code to AI consumes expertise rather than dispensing with it, starting with the junior pipeline that produces it. None is required to follow the core sequence; each rewards the reader who wants the edges of the subject. When the reading is done, [Remake](#remake-assets) holds the mechanisms that turn the thinking into apparatus for a Board agenda, and [the questions](#faqs) are the ones I would put to any executive team proposing workforce AI investment. If you can answer them comfortably, you are ahead of most organisations I see. ## Core reading The sequence runs from the divide now opening in the workforce, through capability building and redeployment, to the future of hours, pay, and the teams that deliver it. 1. [The AI Talent Bifurcation: Are You Building Skills or Collecting Credentials?](https://mariothomas.com/blog/ai-workforce-bifurcation/) (8 minute read, 18 January 2026): Workers with real AI capability command premiums of 28-56%; those collecting credentials without it face a 29% penalty. The same split now reaches the Boardroom. Podcast edition: 12 minute listen. 2. [Upskilling for the AI Era: Building a Future-Ready Workforce](https://mariothomas.com/blog/workforce-upskilling-ai-era/) (15 minute read, 14 April 2025): The scramble for AI talent repeats the early cloud years, and the answer is the same: upskill the people already there instead of chasing hires. 3. [The Redeployment Dividend: Why AI Will Unleash Your People, Not Replace Them](https://mariothomas.com/blog/ai-redeployment-dividend/) (9 minute read, 11 January 2026): AI's primary value is not headcount reduction but the intellectual capital it releases from undifferentiated work. Measure only the former and the dividend goes unclaimed. Podcast edition: 12 minute listen. 4. [Governing the Redeployment Dividend: Turning Saved Hours Into Value](https://mariothomas.com/blog/governing-the-redeployment-dividend/) (10 minute read, 19 July 2026): AI is saving time almost everywhere. The organisations that gain from it are the ones whose Boards decide what the recovered capacity becomes. Podcast edition: 12 minute listen. 5. [Dawn of the three-hour work week: AI's impact on employment and compensation](https://mariothomas.com/blog/dawn-of-the-three-hour-work-week/) (10 minute read, 9 June 2024): If machines do routine work faster and cheaper, what happens to us? The outcome is negotiated, not predetermined. 6. [The future of AI expertise: Building and managing AI-capable teams](https://mariothomas.com/blog/future-of-ai-expertise/) (7 minute read, 21 January 2025): AI's promise of productivity and innovation is delivered by teams, not tools. Building AI-capable teams draws on what the Cloud Centre of Excellence taught me. ## Further reading - [The Personal Agent Economy: When Your Best AI Isn't On Your Balance Sheet](https://mariothomas.com/blog/personal-agent-economy/) (8 minute read, 15 February 2026): Individuals now own AI agents that encode their judgement and expertise, capability that belongs to them rather than their employer. The assumption has inverted. Podcast edition: 10 minute listen. - [The Great Remaking: How the Four Dimensions of Work Are Transforming](https://mariothomas.com/blog/the-great-remaking-four-dimensions/) (15 minute read, 8 March 2026): AI is remaking thinking, deciding, creating and delivering at different speeds and towards different ends. Treating them as one question is most organisations' mistake. Podcast edition: 20 minute listen. - [Europe's AI challenge: Why culture trumps capital in technology adoption](https://mariothomas.com/blog/europe-ai-challenge/) (7 minute read, 13 December 2024): Europe's lag in AI adoption is not a shortage of capital. Seen from San Francisco, the difference is culture: openness to transformation, not money. - [The Accountability Gap: When AI Delegation Meets Human Responsibility](https://mariothomas.com/blog/ai-agency-accountability/) (15 minute read, 16 November 2025): Organisations are transferring decision-making agency to AI while accountability stays with people, and approving deployments without the verification capability that accountability needs. - [The Balancing Item: The AI Oversight Cost Your Business Case Never Priced](https://mariothomas.com/blog/unpriced-cost-ai-oversight/) (10 minute read, 26 July 2026): Every AI business case counts the hours saved. Almost none counts the oversight hours added, and people are silently absorbing the difference. Podcast edition: 13 minute listen. - [The Verification Premium: What Classical Training Reveals About AI Coding Costs](https://mariothomas.com/blog/vibe-coding-vs-classical-training/) (13 minute read, 25 January 2026): AI coding tools amplify the expertise gap rather than closing it: senior developers capture twice the gains. The verification premium is the cost nobody budgets. Podcast edition: 18 minute listen. ## Remake The mechanisms beneath the thinking: the model, diagnostic, methodology, and principle from the Remake Library that turn this briefing into apparatus a Board can use. - **Model: Redeployment Dividend**. The strategic value released when the hours AI frees are redirected to differentiated work rather than eliminated; the headline line of any remaking's business case. [Remake Library](https://mariothomas.com/remake/library/#redeployment-dividend) - **Diagnostic: Process Audit**. The per-process evaluation applied one process at a time: what the work is for, how it is done today and whether it is still needed, and why it needs the technology at all, returning a verdict on the Stop, Keep, Remake scale with every reading graded by indicator type. [Remake Library](https://mariothomas.com/remake/library/#process-audit) - **Methodology: AI Business Case**. The integrated decision framework that crystallises across an ADAPT engagement rather than at a single stage: strategic alignment established at Align, cost and readiness evidenced at Diagnose, value shaped at Advise, and execution designed at Plan. [Remake Library](https://mariothomas.com/remake/library/#ai-business-case) - **Principle: Well-Advised**. The framework of five strategic priorities, Innovation, Customer Value, Operational Excellence, Responsible Transformation, and Revenue, used to ensure AI investments create balanced value rather than narrow cost reduction. [Remake Library](https://mariothomas.com/remake/library/well-advised/) ## Questions The questions I would want answered before approving any workforce AI investment, drawn from the arguments across this briefing. ### Should headcount reduction be the success metric for our AI investment? No, and defaulting to it is the most common mistake I see. Headcount reduction is measurable and immediate, but it treats AI as a cost-reduction tool when the larger value is capability multiplication: releasing the organisation's best people from undifferentiated work and redirecting them toward the judgement, relationships, and innovation that AI cannot replicate. The better measure is where freed capacity flows. I make the full case in [The Redeployment Dividend](/blog/ai-redeployment-dividend/). ### How do we tell whether our AI training builds capability or just credentials? Look at what the spend develops. Premium-earning capability means people can verify AI outputs against domain knowledge, redesign workflows, and handle the exceptions automation fails on. Penalty-suffering credentials mean people can generate outputs they cannot assess. BCG's 2025 AI at Work report found 67% of employees saving over an hour a day where workflows were redesigned around AI, against 49% where tools were dropped into unchanged processes. The test is demonstrated judgement, not course completion. See [The AI Talent Bifurcation](/blog/ai-workforce-bifurcation/). ### Will our workforce resist AI adoption? The evidence says the workforce is readier for AI than the resistance assumption allows. Stanford's 2025 research found 69% of workers welcome automation that frees their time for higher-value work, and Deloitte's 2025 research found workers across all age groups prefer mixed human-AI collaboration. Resistance appears when people conclude adoption is designed to eliminate them; employees who believe it is designed to elevate their contribution become advocates. The transition approach, covered in [The Redeployment Dividend](/blog/ai-redeployment-dividend/) and [Upskilling for the AI Era](/blog/workforce-upskilling-ai-era/), determines which the organisation gets. ### Where do our senior experts come from if AI does the junior work? That is the question most business cases never answer. Expertise develops by doing the work repeatedly, not by reviewing AI outputs, so replacing junior roles wholesale destroys the pipeline that produces the senior experts the organisation will need to verify AI at scale. Augmentation captures comparable efficiency while preserving the pipeline. The strategic choice, and its measurable consequences, is set out in [The Accountability Gap](/blog/ai-agency-accountability/). ### Does the capability question apply to the Board itself? Directly. The IoD's 2026 NEDs Reimagined paper positions AI competence as a core NED responsibility, and its conclusion deserves attention: directors unable to use AI in their own Boardroom work are unlikely to be effective change agents for it elsewhere. A Board that cannot independently evaluate AI strategy ends up ratifying management narratives it cannot challenge, which is oversight in name only. The Boardroom mirror is examined in [The AI Talent Bifurcation](/blog/ai-workforce-bifurcation/). ### Who owns what happens to the hours AI frees? In most of the organisations I see, nobody does. Gartner's 2025 research found teams using AI save around five hours per person per week, most of it draining into non-value-added work, and a 2026 NBER study of nearly 6,000 executives found roughly nine in ten reporting no measurable productivity impact at their own firm. The Redeployment Dividend model pays out only when freed capacity has an owner and a destination: AI owns execution, managers own the workflow, and the Board owns whether the capacity creates value. A business case claiming a time saving without the operating-model change that converts it is incomplete. See [Governing the Redeployment Dividend](/blog/governing-the-redeployment-dividend/). ### Who absorbs the oversight cost our AI business case never priced? In most of the business cases I see, nobody is named, and the cost is paid in attention. The hours saved are counted; the reviewing, correcting, and supervising that AI outputs demand are not, so the work lands on existing people on top of existing jobs. BCG Henderson Institute research published in 2026 surveyed 1,488 US workers and found 14% of AI users reporting mental fatigue, the strain attaching to oversight load, not AI use itself. That is a Board control cost, not a wellbeing issue. The credible answer names the roles, states the share of their capacity oversight consumes, and prices it fully loaded. See [The Balancing Item](/blog/unpriced-cost-ai-oversight/). ### What changes when our best people own better AI than we provide? The employment relationship starts to invert, though the article is explicit that it describes a trajectory, not a timetable. A personal agent trained over a career encodes judgement and expertise that belong to the individual, travel with them, and may outperform anything the organisation provisions. Talent strategy then faces a choice between banning the agents and losing the people, or negotiating access to infrastructure the organisation does not control. The Verification Premium model gains a new dimension when the most capable workers also own the most capable tools, and retention becomes a question of what the organisation offers beyond what people already carry. See [The Personal Agent Economy](/blog/personal-agent-economy/). ## References The research behind the arguments in this briefing, for directors who want the evidence at first hand. - **PwC** (2025): [The Fearless Future: 2025 Global AI Jobs Barometer](https://www.pwc.com/gx/en/issues/artificial-intelligence/job-barometer/2025/report.pdf). The 2025 Global AI Jobs Barometer: a 56% wage premium for AI skills and three times the revenue growth per employee in the most AI-exposed industries. - **Harvard Gazette** (16 September 2025): [AI took your job — can retraining help?](https://news.harvard.edu/gazette/story/2025/09/ai-took-your-job-can-retraining-help/). The retraining evidence: 25-40% of roles are AI-retrainable, but token retraining carries a 29% earnings penalty. - **Lightcast** (2024): [Beyond The Buzz: Developing the AI Skills Employers Actually Need](https://lightcast.io/resources/research/beyond-the-buzz-developing-the-ai-skills-employers-actually-need). Analysis of 1.3 billion job postings quantifying the AI skills premium and its spread far beyond IT roles. - **Anthropic** (2 October 2025): [Introducing the Anthropic Economic Index](https://www.anthropic.com/news/the-anthropic-economic-index). Usage data distinguishing augmentation from automation patterns in real AI interactions. - **Institute of Directors** (January 2026): [NEDs Reimagined](https://www.iod.com/app/uploads/2026/01/FINAL-IoD-Business-Paper-NEDs-reimagined-14.01-6ca5096ee6348f2301347e942a1ffe29.pdf). The IoD business paper whose Recommendation 11 positions AI competence as a core non-executive director responsibility. - **BCG** (26 June 2025): [AI at Work: Momentum Builds, but Gaps Remain](https://www.bcg.com/publications/2025/ai-at-work-momentum-builds-but-gaps-remain). BCG's AI at Work 2025: 67% of employees saving over an hour a day where workflows were redesigned around AI, against 49% where tools were dropped into unchanged processes. - **Deloitte** (24 August 2025): [AI, demographic shifts, and agility: Preparing for the next workforce evolution](https://www.deloitte.com/us/en/insights/topics/talent/strategies-for-workforce-evolution.html). Workforce evolution research showing workers across all age groups prefer mixed human-AI collaboration. - **Stanford Report** (7 July 2025): [What workers really want from AI](https://news.stanford.edu/stories/2025/07/what-workers-really-want-from-ai). Worker sentiment research: 69% welcome automation that frees their time for higher-value work. - **MIT Media Lab** (10 June 2025): [Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task](https://www.media.mit.edu/publications/your-brain-on-chatgpt/). The cognitive atrophy evidence: why excessive AI reliance erodes the critical thinking that redeployment must preserve. - **Harvard Business Review** (16 September 2025): [The Perils of Using AI to Replace Entry-Level Jobs](https://hbr.org/2025/09/the-perils-of-using-ai-to-replace-entry-level-jobs). Payroll-data evidence of a 13% decline in AI-exposed entry-level roles, and the expertise pipeline risk it signals. - **Gartner** (9 December 2025): [Forget Layoffs: AI Is Coming for Inefficiency, Not People](https://www.gartner.com/en/articles/ai-is-coming-for-inefficiency). Gartner's article re-presentation of its 724-respondent productivity survey: AI-driven time savings average five hours per person per week, mostly reabsorbed by non-value-added work. - **NBER** (2026): [Firm Data on AI](https://www.nber.org/papers/w34836). NBER working paper 34836 (Yotzov, Barrero, Bloom et al., February 2026): survey of nearly 6,000 executives across the US, UK, Germany, and Australia on AI adoption and its effects on jobs, productivity, and output. - **Harvard Business Review** (1 March 2026): [When Using AI Leads to “Brain Fry”](https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry). BCG Henderson Institute research in HBR: a survey of 1,488 full-time US workers identifying mental fatigue that attaches to heavy AI oversight loads, with replacement-pattern AI use associated with lower burnout. - **PwC** (13 April 2026): [Three-quarters of AI’s economic gains are being captured by just 20% of companies – with the leading companies focused on growth, not just productivity](https://www.pwc.com/gx/en/news-room/press-releases/2026/pwc-2026-ai-performance-study.html). PwC's 2026 AI Performance Study: around three quarters of AI's economic gains captured by 20% of companies, the concentration the redeployment argument answers. ## The ideas beneath this briefing Ideas I've named and matured writing about AI & the Workforce: what each one means, and where it started. - **AI Capability Bifurcation**: The split between workers who build genuine capability to verify and apply judgement to AI outputs, who command a premium, and those who merely accumulate credentials or tool exposure, who face an earnings penalty. For a Board, the same divide runs through the boardroom itself: familiarity with AI tools is not the same as the capability to challenge what the organisation actually does with them. [Read more](https://mariothomas.com/blog/ai-workforce-bifurcation/) - **Selective Atrophy**: The intentional acceptance that some skills, like routine verification, may safely decline, while strategic reasoning, relationship building and complex judgement must be deliberately preserved through continued human engagement. [Read more](https://mariothomas.com/blog/ai-redeployment-dividend/) - **AI Skills Paradox**: AI simultaneously threatens to automate certain roles while creating acute talent shortages in others, requiring organisations to prepare workers for both displacement and new opportunities at once. [Read more](https://mariothomas.com/blog/future-of-ai-expertise/) - **Human Residual**: The durable human contribution that survives in each dimension of work as AI advances: judgement in thinking, accountability in deciding, taste and originality in creating, and adaptability and trust in delivering. [Read more](https://mariothomas.com/blog/the-great-remaking-four-dimensions/) - **Accountability Gap**: When an organisation delegates work to AI without building the capability to verify it, leaving people answerable for outputs no one has actually checked. For a Board, no delegation to AI should be approved without also approving who checks the output and how, because accountability without a verification step is accountability in name only. [Read more](https://mariothomas.com/blog/ai-agency-accountability/) All concepts: https://mariothomas.com/glossary/concepts/ ## More Board Briefings More complete resources on AI and emerging technology for the Boards that need the full picture. - [The Great Remaking](https://mariothomas.com/briefings/the-great-remaking/): AI is restructuring how organisations think, decide, create, and deliver, and the gap between those that redesign work and those that wait compounds. - [AI Transformation](https://mariothomas.com/briefings/ai-transformation/): Most organisations are stuck at the pilot stage; crossing the divide is a Board judgement about delegation, architecture, and governance, not a technology purchase. - [Operating AI](https://mariothomas.com/briefings/operating-ai/): Most AI pilots never reach production; the AI Centre of Excellence is the operating capability that turns scattered experiments into governed, scaled adoption. - [AI Accountability](https://mariothomas.com/briefings/ai-accountability/): Agency can move to the machine; accountability cannot, and answering for what AI decides now takes capability that policy alone does not supply.