AI, Northern Ireland 2025

The Future

First released in November 2022, Chat GPT has become the fastest growing SaaS tool in history. Back then, GPT was just pretty cool; you could ask it questions and it would write back to you. Cool, but not that useful.

In the (almost) three years since then, the pace of development has been astounding. In another three years from now, what will the core capabilities look like?

Market forces mandate a continued breakneck speed of development. Open AI, Meta, Google, Anthropic, X — are all in a devastating race to build super intelligence. Throw in the requirements for the Western World to outpace China in AI development, and you have a race scenario.

Any window for a moratorium on AI safety has passed. The genie is out of the bottle.

What kind of roles will be impacted, and why?

Job displacement will happen gradually, and then suddenly. AI won’t neatly slide into organisations and perform as well as humans — better to think of it in terms of tasks; what tasks can AI now perform a human level?

Market research, analysis, content creation, basic administration and other workplace staples will be subsumed by AI; this will erode the core responsibilities of a lot of job titles:

(i) Paralegal. AI can train on the entirety of a legal system in seconds. Parsing, reading and creating legal documentation will be AI led.
(ii) Marketing. AI is now simply better at image generation and content generation than humans.
(iii) Business Development. Activities like lead generation and pitch deck creation can now entirely be automated. Don’t discount the buy side, either, which will be getting more sophisticated at finding the right product or service.
(iv) Analyst. It doesn’t really matter what you are analysing. Finance, consumer data or hippopotamuses. AI does it better, faster and cheaper.
(v) Web Developer. Tools like Replit and Loveable are rendering web developers increasingly redundant; prompt engineering generates code that be deployed in a live environment in seconds.

More broadly for AI disruption, look for high margin sectors of the economy like Management Consulting and Asset Management. The free market won’t support those markups much longer.

What kind of roles are future proof?

The “safest” knowledge roles share four traits:

(i) Leadership — do you lead a team and own strategy?
(ii) Nuance — is there complexity that only you can manage?
(iii) Human interaction — do you engage with multiple stakeholders?
(iv) Institutional context — are you woven into the history and fabric of your organisation?

These traits might map across to cross-functional project management, stakeholder management, problem solving or public speaking. Think of leadership roles, operations and client facing delivery roles.

Beyond that, roles within AI architecture or delivery should remain in high demand and command premium salaries.

Won’t AI create new roles?

“Luddite” is a term used to describe someone as old fashioned. A Luddite was, in the pre-industrial revolution era, a highly skilled hand weaver. They were replaced by machinery as the UK’s economy gradually transitioned away from muscle power to knowledge. That was traumatic in the short term and caused volatility, but over time, they found new employment and the economy adapted to the advent of machinery.

It is possible that the emergence of AI is similar to the industrial revolution; a natural evolution that will cause initial turbulence before returning to a steady state of labour occupation — but I can’t necessarily think what those roles might be beyond the obvious AI Auditors and integration specialists.

What public policy interventions can be effective?

Northern Ireland has world‑class cyber capability at CSIT in Belfast; real‑time analytics leadership via FD Technologies in Newry; a deep tech employer base at Allstate NI; and advanced manufacturing at Seagate in Derry. Blue Canvas offers direct AI consultancy. Project Kelvin has provided have direct transatlantic fibre and the AICC is already providing skills frameworks.

Northern Ireland is better placed than most and has the opportunity to adapt rapidly to the era of superintelligence and become a model economy for others to emulate.

• Special Economic Zones. AI infrastructure is an extremely valuable commodity. Find a vacant plot of land, set up a SEZ and make overtures to one of the hyperscalers; Meta, Google, Open AI, whoever — and let them build a huge data centre. It is likely that this will become a dominant source of tax income going forward. That’s your leverage.
• Micro enterprises. AI will drive a significantly enhanced return on deployed capital; £50k of seed capital might allow a startup to tool up, build and become cashflow positive within 3 months. Back startups — they will be able to do a lot more, leaner and faster.
• Exports > Job Creation. Job creation will no longer be a suitable yard stick for assessing economic health. Dialling in on Northern Ireland’s export potential is therefore a shrewd move; anything that promotes capital flowing into, not around, Northern Ireland is preferrable.
• Darwinsm. The health of internal markets such as accountancy is a zero-sum game; there is a finite demand for services and AI first mover advantage will mean that some will flourish and get the market share, while others languish. If a legal firm adopts AI and reduces their costs by 40%, this will allow them to conveyance properties for 30% less than market rate and they will own a market — but that won’t increase GDP since late adopters will fail. Employment rates will fall regardless, so economic interventions wold be inefficient.
• Knowledge Transfer. In most first jobs post-University, completing spreadsheets and taking the minutes are the norm. Adding value isn’t really expected of recent graduates — organisations are essentially investing in graduates in the hope that they can become value add professionals. Knowledge transfer by osmosis with peers is invaluable to development. That will no longer happen organically. Gen Z will not be learning from Boomers. Finding a way of faciliating knowledge transfer from senior to junior and plugging that economic skills gap is critical.
• Empower People. AI is not democratized. The average person does not pay for a premium tier of AI. Pro-tier Open AI users can access Deep Research, Agents and other models that the average Joe cannot. Northern Ireland could consider taking out an enterprise license of AI and giving qualifying citizens access to this.

There are no easy answers — no silver bullets, but rapidly deploying a range of measures and testing their impact is advised. AI isn’t slowing down; stasis, in this instance, is much worse than making some proactive missteps:

Nightingale Hospitals

We’ve had the doom and gloom; what of the good news? Well, it might just be possible that if AI achieves superintelligence, it could usher in an era of human abundance. It might cure or prevent cancer, it might solve the climate change crisis with renewable energies. It could tackle all of humanities challenges and maybe we all live fulfilling, optimised lives as a result.

In March 2020 as the scale of the Covid epidemic begin to register with policy makers, the UK ordered the creation of Nightingale Hospitals to deal with patients in the event of NHS infrastructure being overwhelmed. Over time, lockdowns kept the impacts of the virus to manageable levels and the Nightingales were seldom used. They were scorned and ridiculed as a needless expense.

But, we would have been glad to have them if they were required. They were a proactive failsafe during uncertainty. Policy actions in the era of AI feels a little like Nightingale Hospitals; not definitely required but extremely useful if they are.

Caution and stasis in the age of AI are akin to economic death knells. Time to be bold.

‍