Artificial intelligence is no longer emerging; it is already embedded in how society functions. It shapes how information is ranked, what people see online, how decisions are made and, increasingly, how power is exercised. This reality requires us to move from abstract discussions and act with urgency, clarity and competence. One of the biggest misconceptions about AI is that it is something people choose to use. In fact, the most influential AI systems today are the ones people never consciously engage with. They sit inside platforms, search engines, social media feeds, recommender systems and automated decision-making tools that quietly influence behaviour at scale. This is where the real impact is. System such as Grok, which we have discussed this week, embedded directly in social media platforms demonstrate how quickly AI can shape public discourse when it is trained on live data, optimised for engagement and applied in real-time and, in Grok's case, enables legal non-consensual image manipulation. These systems are not neutral. They reflect incentives, values and commercial priorities. When these priorities conflict with the truth, well-being or democratic norms, the consequences are serious. Another major shift that is still not widely understood is the move from AI that responds to AI that acts. We now see the rise of autonomous AI agents that will carry out tasks across multiple systems, write and deploy code, interact with other AI tools and make decisions without continuous human interaction. These systems are already operating enterprise environments and will soon appear in public services. Our governance frameworks are not yet equipped for this level of autonomy. We must also confront the impact on children and young people. Algorithms are designed to maximise attention and emotional engagement. For children this can mean exposure to content loops that amplify anxiety, distort self-image and damage mental health. If we fail to recognise algorithmic exposure as a child protection issue, then we are falling behind the reality that families already deal with. This is why the establishment of the national AI office is not only important but urgent. I published legislation to create the office because I believe the State needs dedicated expert capability focused solely on AI. The office must be more than a future structure box ticking exercise. It needs to be operational now. AI systems have already been deployed across public and private sectors. Decisions are already being automated. Data is already being processed at scale. If the national AI office is still finding its feet in eight or ten months' time, we will already be regulating after the fact. What should the office be doing immediately? It must act as a centre of real technical expertise in the State, not only regarding a legal interpretation of things such as the EU AI Act but a genuine understanding of how models are trained, how data flows and how risks emerge in practice. Without this, oversight will always lag behind deployment. The office must actively guide public bodies that are already experimenting with AI. Departments and agencies need clear practical advice on what is acceptable, what is risky and how to deploy AI safely. Silence or delay create uncertainty and drive poor decision-making. The office must be able to intervene quickly. AI deployment cycles are measured in weeks and, in some cases, days rather than years. If our response mechanisms are slow, fragmented or overly cautious, and if we are dealing with AI purely in the political cycles, then we will fall behind. We have already fallen behind and let us be clear about this. AI is not waiting for us to get comfortable with what it is and what it is doing. It is moving fast and it will continue to do so regardless of how prepared we are. The real risk is not that we act too boldly but that we move too slowly and with too little understanding. The natural fear that is there around AI has caused us to act slowly and to double check and double think what we are doing, particularly when it comes to regulation. Particularly when it comes to the national AI office, we need to act quickly and boldly and ensure that people can see the State is working to protect the interests of the public. Ireland has the talent, the credibility and the values to lead in AI, particularly with regard to regulation, but leadership requires urgency, expertise and the willingness to engage with AI as it exists today and not as we might wish to exist in future. We can shape it if we engage. Putting our heads in the sand and asking whether it will go away will not cut the mustard.

Sentiment score: 0.11