Being honest, I did not intend to speak on these amendments - I was going to wait until later - but when a substantial debate, which should go on for a long period in a committee room, is squeezed into two hours, it invariably turns into Second Stage speeches because it is our last chance to have a say on it. For me, this is the third leg of a bonfire. I know bonfires do not have legs but I am referring to whatever the wooden structures are that go around a serious bonfire. This is the third one we have had in one evening as we burn up our climate targets, the other two being much more obvious, namely, the Bills on LNG and the passenger cap in Dublin Airport. However, this is also part of it. As regards the section we are dealing with here, the fundamental problem that some of the amendments, and I am assuming the Minister of State will reject them all, attempt to address is the fact that the AI office is not independent and is subordinate to the Department of enterprise. Even the location of the office, subordinate to the Department of enterprise, tells us everything we need to know. This is then fleshed out in the mandate of the AI office and so on. This tells us what the mindset of the Government is in relation to AI is. It is that this is an enterprise issue that is overwhelmingly viewed through that lens. Its view is that this is a train that is moving or has already moved and we all need to get on board, and given that all the companies seem to be doing it, we need to educate all our young people in using AI. That is the whole mindset. I fundamentally disagree with that, as my colleagues on the committee will know. In general, from where we are currently standing, AI is a force for bad and not good in the world. It is accelerating genocide, climate change and mental health crises, and it is beginning to accelerate significant job loss and displacement. It is eroding workers' autonomy for those who still have jobs, and it is driving misogyny and hate. I do not deny some discrete, positive applications we have heard about for weather forecasting, healthcare and so on, but, as a whole, any objective look at the current impact of this technology will conclude that it is a negative thing. However, what we are seeing in the structure of this Bill, and we will get to the mandate in section 9 shortly, is green for go and doing everything we can to support and promote AI. It very much sees things through an enterprise frame, which is very problematic.

Sentiment score: -0.12

I have two points. One is the point that Deputy Gibney made. The Minister of State denied that she gets to appoint the CEO. For now, until the establishment of the board, she has total power in appointing the CEO. That is the truth. That is what is set out in the legislation. Afterwards, she still gets to pick the CEO. She picks the CEO from a shortlist, which is done by the Public Appointments Service, but she does pick them. The Minister of State may feel free to correct me if I am wrong in that. I am not sure her strong rebuttal of Deputy Gibney really holds up. My second point is that probably the worst thing in this Bill is the inclusion of the mandate of promoting and fostering AI innovation and literacy in the AI office. This is financial regulator stuff, where the financial regulator was promoting finance and simultaneously regulating finance. You cannot be poacher and gamekeeper at the same time. They point in opposite directions. If the Government wants to have a body whose job it is to promote AI literacy and so on, it can knock itself out, but it should not be the same person and office that is supposed to be regulating and ensuring this does not cause harm to society. That is the whole problem here. It will have both the good mandate and the bad mandate set up together in this one office. It is sitting under the auspices of, and not really independent of, the Department of enterprise, which is very much on one side of that equation. That is the fundamental problem. The Minister of State can make the case to have people advocating for fostering AI innovation and literacy and so on, but the idea that the body carrying out that function will also be the regulator is very problematic.

Sentiment score: -0.05

It is appropriate that I speak at this stage because I think that gets to the point. I think the Government agrees with the Deputy that it wants to set up an enabling office. That is the wrong way to go. Fundamentally we need a “thou shalt not” regulating office to try to protect the citizens from the potentially negative impact of AI. I will speak to two of the amendments. I support Deputy Gibney's amendment on adding in the promotion of the responsible use of AI as opposed to the current proposal we have. It is amendment No. 2 and speaks of “ethical and responsible”. I presume we are going to hear that we are not going to get that, but it is really not too much to ask. Second, Deputy Lawlor's amendment No. 7 on the climate impact of AI is extremely important. Last week we had Jennie Stephens and Hannah Daly in. It is funny and I have quoted this to a few people who have been in. KPMG has a report whereby it surveys chief executives of companies and asks what they think about AI and 95% of them say that AI is really going to help us deal with climate change. I do not know why we are asking the chief executives. Let us ask climate scientists if AI is going to help us deal with climate change. I have not found one yet that says yes. Everyone says this is really a problem in terms of climate change.

Sentiment score: 0.05

Does Deputy Geoghegan mean the weather forecasting guy?

Sentiment score: 0.00

His main thing was to say this is not like the large language models, LLMs. I did not disagree with anything he said but I do not think he was making the general case that is being made that somehow AI is going to help us to deal with climate change. If we roll out the argument that a combination of AI and a stable demand for energy is going to promote investment in renewables, and load management and so on will happen at a better scale, I really think it is magical thinking. I do not think we are going to find climate scientists to make that case. Instead, what we heard from Jennie Stephens was that AI is preventing the phase-out of fossil fuels. The rapid deployment of AI has led to a dramatic increase in fossil fuel exploration, extraction and burning. The thing that struck me is that AI is being used to make fossil fuel exploration and extraction more efficient in a sense. Professor Stephens also said, "Tech companies are also collaborating with the fossil fuel industry to develop AI tools used to accelerate fossil fuel exploration and allow fossil fuels to be extracted more quickly and cheaply". She argued "AI promotes climate obstruction" and denial, delay and distraction. I thought this point was very strong: "Tech companies are now deploying many of the same strategies that the fossil fuel industry has used to squash public concern and block climate policy", strategically squashing research, discussion and awareness about the environmental impacts of AI. We also heard from Professor Hannah Daly about the reality many of us have been pointing out for a long time, which is that all the extra renewables we are adding to the grid are just being sucked up by the data centres over an extended period. For me, it did crystallise something and I felt our guests today were trapped in this a bit. Often, there is just the illusion that somehow adding more renewable capacity, in and of itself, is in some way dealing with climate change but it is not. We can add all the renewables that we want but as long as we keep expanding usage we are not getting any closer to reducing greenhouse gas emissions. The climate does not really care if we have a whole load of renewables happening over here but we are still increasing or sustaining the same level of greenhouse gas burning. The amendment is good because it says that this should be central and that the Climate Action and Low Carbon Development Act has to apply here. We have had it disapplied on a number of occasions this evening. That is what we are generally doing this evening, but AI has to be consistent with our supposedly legally binding climate targets.

Sentiment score: 0.07