🎧 Prefer to listen? Audio version below — approximately 9 minutes.
There is a report circulating with Ernst and Young's branding on it. At first glance it looks like the standard output of a major professional services firm -- structured, confident, authoritative. A closer read reveals fabricated citations and invented data. AI detection specialists at GPTZero were unambiguous: the document had all the hallmarks of unsupervised AI generation. A firm that sells governance, assurance and risk management to some of the world's largest organisations had published, under its own name, something it apparently could not be bothered to check.
It is a small incident in the grand sweep of the AI moment. But small incidents have a way of illuminating large problems.
The consulting industry is currently in the grip of a hiring panic. Forward Deployed Engineers -- a job title Palantir invented for the coders it flew into conflict zones to work alongside US intelligence operatives -- are now being recruited by KPMG, Deloitte, PwC and EY. Anthropic is paying up to $300,000 for them. The Big Four are offering $260,000 and losing. Everyone is repositioning as an AI-native operation. The slide decks are compelling.
What the slide decks do not address is the question of who comes next.
The firms are racing to hire elite, high-cost engineers at the top to build systems that automate away the very junior roles that used to produce their future partners. The irony is structural and nobody is saying it out loud.
The junior associate grinding through her third set of management accounts at eleven o'clock on a Tuesday night is not just cheap labour. She is learning something that cannot be delivered by a prompt. She is developing the pattern recognition that tells an experienced practitioner which number looks wrong before they can explain why. She is building the judgement that distinguishes a control weakness from a catastrophic exposure. She is, in the unglamorous language of professional development, becoming someone who actually knows what they are doing.
Hand that work to an AI agent and you have not eliminated the problem. You have deferred it by about a decade, at which point the senior partner retires and the person behind her has spent her career supervising model output rather than understanding what she is looking at. Supervising output and understanding what you are supervising are not the same skill. That distinction will not be obvious until it matters enormously.
This argument applies everywhere. On the Isle of Man it carries a particular weight.
The Island has always hired from outside itself. That is not naivety, it is pragmatism -- a small jurisdiction with a sophisticated regulatory environment and a specialised economic base cannot grow every skill it needs from within its own population. The talent that has come here from the United Kingdom, Ireland, South Africa, Zimbabwe and elsewhere has built institutions, brought expertise, and contributed to a jurisdiction that punches well above its weight.
But there is a difference between supplementing a pipeline and substituting for one. Importing talent is a dependency. It works when the Island is an attractive destination, when the housing market does not price out young professionals, when the salary differential justifies the relocation. None of those conditions are permanent or guaranteed. The Island is already pricing out its own young people before they have built careers, while simultaneously making itself a harder proposition for the mid-tier imported talent it depends on to fill the gaps. Pressure on both ends of the pipeline at once is not a comfortable place to be.
More importantly, none of it addresses the question of whether the Island is enabling its own people to participate in the economy it has built.
The entry-level work that AI is now displacing was never just about filling seats cheaply. In financial services, egaming compliance, legal practice -- it was the mechanism by which a young person from Peel or Ramsey could spend a decade becoming someone with genuine expertise and a stake in the place. Remove that mechanism and you have not solved a resourcing problem. You have closed an on-ramp.
The regulatory environment remains demanding regardless. The GSC, the IPA, the FSA require real competence, not supervised model output dressed up as professional judgement. That competence has to come from somewhere. If it comes entirely from outside, the Island has quietly outsourced its human capital formation to other jurisdictions' education and training systems -- a dependency that sits in the same category as routing your critical data infrastructure through a US hyperscaler. Convenient, functional, and entirely outside your control.
The question is not whether the Island can hire from outside. It demonstrably can. The question is whether it should be doing more to ensure that the next generation of professionals includes people who grew up here, chose to stay, and built their careers in a system that gave them the room to develop. AI adoption that removes the lower rungs of that ladder is not progress. It is a policy choice being made by default, in individual hiring decisions and procurement conversations, without anyone explicitly owning the consequence.
Here is where the productivity narrative becomes dishonest.
When a professional services firm automates the work that junior staff used to do, the productivity gains are captured by the firm. The partners earn more. The shareholders -- where there are shareholders -- do better. The slide deck about digital transformation gets another bullet point.
The costs go elsewhere.
No entry-level work means no career ladder. No career ladder means a generation for whom the professional sectors that have historically offered structured progression simply do not have room. On an island with a specific economic structure and limited alternatives, that is not an abstraction. That is people on benefits. That is income tax receipts that do not materialise. That is National Insurance contributions that are never made. That is a public finances calculation that gets quietly worse every year while the productivity statistics look encouraging.
And it does not stop there. Worklessness is not just an economic condition. The evidence on long-term unemployment and underemployment is not ambiguous -- poor mental health outcomes, reduced life expectancy, increased demand on health and social services. The Isle of Man's health infrastructure is under pressure that is already visible. Loading additional demand onto it, funded by a shrinking contributory base, is not a technology question. It is a policy choice that is currently being made by default, in boardrooms, without anyone explicitly deciding it.
None of this is an argument against artificial intelligence. The tools are genuinely useful. Used well, they make experienced people more productive, surface information faster, reduce the friction in complex processes. That is real value and it would be foolish to pretend otherwise.
The question is not the technology. The question is the choices made around it.
There is a difference between using AI to leverage the capabilities of people who know what they are doing, and using AI to avoid the expense and inconvenience of developing people who know what they are doing. The first is productivity. The second is deficit spending on human capital -- borrowing against a future where the expertise still exists, and quietly spending down the principal.
The bill for that kind of borrowing does not arrive immediately. It arrives when the senior partner retires. It arrives when the regulator asks a question that nobody in the room can answer from first principles. It arrives when the Island needs the next generation of professionals and discovers it forgot to grow them.
It arrives, in other words, precisely when you can least afford it.
The EY report with the fabricated citations is a small thing. But it points at something large: the gap between performing expertise and possessing it. That gap can be papered over for a surprisingly long time. It cannot be papered over forever.
Be careful what you wish for. The productivity is real. So is the bill.
The Sovereign Auditor covers digital sovereignty, cybersecurity governance, and data protection policy—with particular focus on Isle of Man jurisdiction and Crown Dependency issues.
Payments via PayPal. Credentials delivered by email. No Substack. No Stripe. No middlemen.
