A decade of digital leadership
Over his 10-year spell as Government Chief Information Officer (CIO), Barry Lowry has emerged as one of the most distinguished civil servants in the State. Due to depart the role in April 2026, Lowry speaks to Joshua Murray about the evolution of AI and digital services over his time in office, the threats and misconceptions about AI, and his plans beyond OGCIO.
When Barry Lowry arrived in Dublin in 2016 to take up the post of Government CIO, Ireland’s digital landscape was at an inflection point. Momentum had stalled, major ICT programmes faced scrutiny, and confidence required rebuilding.
A decade later, Ireland ranks strongly in OECD and World Bank digital government assessments. Core platforms such as MyGovID are embedded, life events reform is reshaping service delivery, a digital wallet is about to go into beta launch, and AI has moved from exploration to structured deployment across the public sector.
“For the last 10 years, we have done exceptionally well,” Lowry says. “Ireland is now right up there in the OECD ratings, right up there in the World Bank ratings. It has been an incredible achievement for a country with a relatively small budget for digital transformation.”
From proof of concept to policy
Lowry is deliberate when describing the trajectory of AI in the public service. “I think ‘evolves’ is a good word,” he says.
The Office of the Government CIO (OGCIO) undertook three proofs of concept spanning front-office and back-office applications. Some departments conducted their own pilots, while OGCIO projects examined both structured and unstructured data, alongside chatbot functionality.
“What we have seen is probably an acceleration of use of AI, but not anybody going over the top in terms of pace,” Lowry says. “It is cautious enough, but it is allowing innovation to happen and rollout to take place.”
Within his own department, all staff have access to Microsoft Copilot. Compared to earlier experimentation, Lowry says current tools are “better, cheaper, and more agile”.
“Copilot is effectively supporting the user in curating the information there and then for the query that they have,” he explains. “As you would expect, people who naturally lean towards use of technology are doing really great stuff with it. Those that are not that interested are keeping it at arm’s length. It is no different to any introduction of technology.”
On adoption of the use of AI, Lowry asserts that this follows familiar patterns, with enthusiasts innovating early, while some prefer to wait while confidence in the technology grows over time.
“I would say the evolution has been very positive,” he says. “That trajectory will continue over the next two to three years. Confidence will grow. Capability will grow and you will see a lot more happening.”
“Leaders should remove obstacles rather than micromanage. The idea is you get things out of the way for them.”
Gov.ie
The most visible development is imminent, with the launch of a chatbot on gov.ie. “We are about to launch a bot on gov.ie. We are just working through all the preparations.”
The system draws on lessons from earlier pilots but is underpinned by a strict design principle in that it uses only validated gov.ie information.
“It is using stuff that we know is valid, and we can stand over,” he explains. “It is giving people a better front end to access that information and a better search facility to find things.”
The launch will be in beta format. “That is the whole point of going live in beta,” he says. “You learn what people actually want.”
The same philosophy underpins work on Ireland’s digital wallet, which is currently in final testing. Feedback loops, including short post-transaction surveys and even video submissions that can be analysed and summarised using AI, are being explored.
“What we want is for people to choose the online route,” Lowry says. “That frees up resources to help the most vulnerable. But if you start to drop off [a digital transaction], why did you drop off? Was it something we did? Or did you just run out of time?”
Responsibility and accountability
As use of AI has become more widespread, questions of accountability have sharpened. Lowry explains: “Ultimately, it is the practitioner that is responsible. They are given excellent guidance, aids, and tools and so there is no excuse for irresponsibility.”
Referring to some well-publicised incidents in other countries, where AI-generated reports were used without checking references, Lowry states that professional accountability must take precedence over the mere use of the technology.
He analogises: “If you were writing an article and you deliberately used false information, is that on you or on your employer? Ultimately, your employer’s name gets dragged through the mud, but it would still be your lapse.”
The Irish Government’s AI guardrails therefore emphasise that humans remain accountable for everything they submit into public discourse or ministerial decision-making.
“Nobody should be submitting anything to a minister or into public dialogue that has not been double and triple checked. They know this and they are trained in it
“Anybody can make a mistake but if you are deliberately not checking material that you are submitting, that is inexcusable.”
Why AI gets things wrong
In reference to AI gone wrong incidents, Lowry is asked why AI generates false content. He states that is “all to do with the data”. He explains that, when asked a question or told to carry out a task, an AI system will attempt to find evidence and construct an answer based on that material. He also says that it does not fabricate with intent, but rather synthesises from available sources.
“The bot will not lie,” he says. “It finds one source of material or joins to one source of material, which happens to be wrong.”
The fluency of the bot’s output, however, can create ambiguity when false or inaccurate information is used. “You always look at the references,” he says. “If I ask a bot to carry out a task for me and I do not like the references, I will not use the answer.” He compares the bot’s actions to a child offering the best answer it can with the knowledge it has.
High-stakes contexts
The principle of verification becomes more acute in frontline contexts. In healthcare, AI has the ability to assist clinicians in analysing information. However, unchecked error would have profound consequences.
“If you have got a cancer diagnosis wrong because you did not check the information properly, that is catastrophic,” he says. Consequently, clinicians will use AI to supplement other tests such as physical examinations and blood tests.
Similarly, he references the debate around facial recognition technology in policing and public order contexts. In Italy, facial recognition has been tested in football stadiums to identify individuals engaging in racist abuse.
Such applications are intended to protect victims and deter misbehaviour, but they also raise privacy concerns. Lowry analogises that this exemplifies “the moral debate every country needs to have”. “AI will not do this by itself. It will only do it if policy has enabled it.”
Regulation
Lowry believes regulation is necessary, particularly given the rise in online scams and exploitation of vulnerable people.
“I think most people, certainly in Ireland, would say we need some regulation,” he says. “It cannot be a totally free game for everyone, because that is where abuse and fraud kick in.”
At the same time, however, he cautions against overcorrection: “It is just about getting it right,” he says. “It is not the technical solution that is hard; it is getting the policy and getting the consultation right.”
Agentic AI and service orchestration
Looking ahead, Lowry identifies two ways in which AI will underpin future public services. The first is advisory. As government reorganises services around life events such as starting a family, high-quality content becomes central.
“If that content is really good, the advice will be really good. It gives people assurance. It avoids them, for example, going to a government office or GP unnecessarily. The system works quicker and is unclogged.”
The second and, in Lowry’s view, more transformative, is orchestration through agentic AI. “The whole idea of agentic AI is it sets up this menu of services and creates an experience that is more than the sum of the parts,” he explains.
Lowry adds that different specialist bots could interact across departments such as Health, the HSE, Social Protection, and Revenue to deliver a coordinated experience for citizens navigating a life event. “It brings in different solutions from different parts of government,” he explains.
“Effectively, how it works is that you have a trigger event, for example the birth of a child, and with the mother’s permission, related services such as birth certificates, vaccine appointments, and benefits are initiated automatically.”
Job losses
Despite widespread commentary about automation and job losses, Lowry does not foresee widespread job displacement due to AI.
When asked if AI will lead to job losses, he says: “Absolutely not. AI cannot ultimately look after customers well and what we are very, very good at in Ireland is looking after people.”
Instead, he anticipates a shift in emphasis. “We will see less time on forms and related back-office activities, and more time looking after people.”
Next steps
Lowry’s path to digital leadership was far from predetermined. Growing up and working for most of his professional life in the North, he initially considered becoming a teacher. A summer job in the Northern Ireland Civil Service, however, changed that trajectory.
Encouraged by his mentor to apply for a trainee programmer role, despite professing that he ‘did not like’ computers, he entered the world of IT. He studied part-time at Queen’s University Belfast and progressed through various analytical and leadership roles, eventually becoming the equivalent of CIO for the Northern Ireland Executive.
After nearly a decade in that post, he received a phone call encouraging him to apply for Ireland’s Government CIO role. “I applied, and the rest is history,” he says.
Asked what advice he would offer his successor, Lowry says: “When you are in a leadership role, my motto has always been ‘help brilliant people be brilliant’.
“Leaders should remove obstacles rather than micromanage. The idea is you get things out of the way for them.”
Concluding, Lowry says: “The best advice you can give anyone about successful transformation comes from an old friend of mine and is honour the past and invite people to co-create the future.”
