On Thursday, the Government launched guidelines for the responsible use of artificial intelligence in the public sector.
It highlighted the dos and don’ts of AI use by Government departments and State bodies, and also revealed various AI projects that are already under way within the public sector.
Unions gave a broad welcome to the plan, but said it did not focus enough on the potential job losses that could result from the widespread adoption of AI.
AI that is permitted
The guidelines state that public sector employees can use AI algorithms and models to generate draft versions of documents, summarise reports and assist with translations.
It is also permitted to use the technology to create new content such as text, images, audio or video based on a set of input data, parameters, or prompts.
It can be deployed to identify and categorise data, analyse documents, detect anomalies and for disease diagnosis in the health sector.
The plan outlines the importance of having safeguards and human oversight in place and the need for AI systems to be dependable and secure.
The guidelines state that systems must have robust policies to protect personal data and highlights the need for the public service to be transparent and upfront with end-users when AI systems are being used.
The plan also states that AI systems that are being used must be fair and inclusive.
“AI is quite fragmented in the public sector right now but these guidelines are designed to encourage and empower public servants to broaden its use across the board,” said Minister for Public Expenditure Jack Chambers.
Examples of AI already in use
The guidelines contain examples of how AI is already being used across public services, including at St Vincent’s University Hospital in Dublin where staff are exploring the potential for AI to assist with performing heart ultrasound scans, in order to help reduce waiting times for patients.
The Revenue Commissioners are using large language models to route taxpayer queries more efficiently.
The Department of Agriculture, Food and the Marine is developing an AI-supported solution to detect errors in grant applications and reduce processing times.
It means that while farmers break ground in their fields, groundbreaking technology is being used to process their information.
Head of data management and analytics in the Department of Agriculture Dr Catherine Dempsey said the AI system they use prevents potential data breaches of customer information and ensures that applications can be processed quickly.
“Where before this was a manual process where an individual would have to step in for an identification that may take weeks, now we are down to minutes and even seconds to process, isolate and identify potential breaches,” Dr Dempsey said.
“That automation means we have increased the efficiency of identifying and rectifying potential breaches,” she said.
One of the central recommendations contained in the guidelines is that there must be human oversight of AI systems that are used within the public service.
It is something Dr Dempsey said they are very conscious of.
“There is constant monitoring and evaluation to ensure that the way the AI is working is the way we intended,” she said.
Free GenAI models not to be used
The framework advises against the use of free generative AI (GenAI) models. This would include popular chatbot tools like ChatGPT and DeepSeek.
“Free GenAI tools are very accessible but because they lack suitable management and oversight pose significant risks for use in the Irish public sector,” the guidelines state.
“Any information given to a public GenAI tool could be used in training the model. Thus, we advise against their use in the public service,” it adds.
Potential future job losses
Unions representing public sector workers gave a broad welcome to the guidelines, but expressed concerns about jobs being displaced by AI.
Fórsa, the country’s largest public service union, said there will be significant challenges to address when the new technology is integrated into the public sector.
“As AI develops it will put various types of jobs at risk, including a variety of skilled jobs,” said Fórsa National Secretary Catherine Keogh.
“As that happens, there needs to be a real worker dividend.
“For example, this opens up opportunities to talk about trialling a four-day week,” Ms Keogh said.
The Association of Higher Civil and Public Servants held its annual conference yesterday and delegates debated a number of motions relating to AI.
Resolutions put forward by delegates highlighted the need for greater management of AI in the workplace and the protection of human decision-making in the public and civil services.
One motion called for the civil and public service to ensure “that human decision-making and performance assessment are retained as exclusively reserved functions of human actors”.
Members also called for a multi-union approach to AI and for the AHCPS to seek commitments from relevant departments to review the governance structure and more specifically the operations of AI within departments that have introduced its operation.
“Like in many workplaces, AI is already playing a positive role in the civil and public sector, but as the technology develops it’s critical that there continues to be strong guidelines and guidance in place,” said AHCPS General Secretary Ciaran Rohan.
“Not only are there issues of confidentiality, bias and transparency to consider, but we would also have concerns about any reliance on ‘algorithms’ to inform decision making,” he added.
Minister Chambers said broadening the use of AI is not intended to reduce numbers in the public service.
“It’s about allowing public servants who might be doing repeated tasks that could be replaced with AI, to be doing different tasks in the context of their work,” Mr Chambers said.
“It’s about partnering with the workers to ensure there is a human interface in all aspects of the AI solution with the core goal of delivering better public services,” he added.
The public sector is strongly unionised, and its workers are in general better protected than those in the private sector.
If there are to be future large-scale layoffs as a result of AI it will likely happen elsewhere first before it hits Government jobs.
Those pushing the roll out of AI say it is about using the technology to replace mundane, repetitive or time-consuming tasks, rather than replacing workers.
The Government insists that our public services will remain human-centred and that using cutting-edge technology doesn’t mean cutting people out of the picture.
