We have seen EirGrid's AI forecasting tool, built with a technology called GridZero.ai, which is able to predict grid conditions 38 hours in advance, and research ongoing that is linked to the StopFloods4.ie project, which applies machine learning to flood risk. These are just a couple of examples of where AI is being used for good and to solve different environmental challenges.
The third part of the ledger is perhaps the most complicated; the system effects. This is how AI changes the broader economic system by making things more efficient, reducing prices and therefore potentially increasing demand. This is known as Jevons paradox or the rebound effect. In a nutshell, the efficiency often generates more consumption and that increased consumption can offset or even reverse the positive benefits I just described. The rebound effect is critical in understanding the trade-offs and this net balance between the positive handprint and the negative footprint.
This all sounds very complex but I am happy to report two new standards from the International Telecommunication Union, L.1480 and L.1801, that are helping governments do this net calculation between all three parts of this ledger. I am also happy to report there is now independent testing of L.1480 by the European Green Digital Coalition and France's environment agency, and real-world test cases are currently being done across 39 use cases. The results from these are fascinating. The enabling benefit for carbon dioxide reduction is orders of magnitude higher than the carbon dioxide footprint. The footprint is a small amount and the enabling mitigating effecting is a much larger one. There are significant positive effects even when the footprint is taken into account. That is what many of the case studies show but some also show the rebound effect can completely wipe out any of the positive environmental gains, so it is fundamental to look at that as well.
Getting the balance right between the footprint, the handprint and the system effects, and ensuring each is measured properly and netted against each other, is the core challenge of Ireland's twin transition in practice. We must also acknowledge these costs and benefits do not land evenly. As mentioned, local communities often carry the footprint while the economic and strategic benefits are national and often reach all the way to the EU or to the international goals. Any credible accounting has to work at all three scales, local, national and transboundary, and not just one.
I will conclude with six concrete steps the committee can take. Right now, the Commission for Regulation of Utilities connection policy only requires conditionality for renewable connections for electricity. This could be expanded to require disclosure of other forms of impact, such as on water, land, minerals and the creation of e-waste.
Therefore, extending beyond renewables to other potential impacts would be important.
The second step would be the application of net benefit calculations for any public environmental funding. Any project that receives State funding, whether it is a grant, procurement contract, development project or capital investment, that claims an environmental or efficiency outcome from AI solutions must substantiate that claim using the ITU-T L.1480 or L.1801 method. This is really meant to avoid the greenwashing we see a lot of, which suggests that AI is going to save the world. No, we need to make sure we are calculating using a standard international methodology and really looking at the potential net positive benefits.
The third intervention to potentially consider is to give the new AI Office of Ireland an explicit mandate for monitoring net environmental benefits from AI for Ireland. I understand the office has just been established as the AI Act's co-ordinating authority for the EU. Potentially, its mandate could be extended to add environmental disclosure and net accounting verification to its remit. If that is not the right place, perhaps there is another institution that could take on this equivalent function.
I have three more recommendations and then I will finish. The fourth recommendation is to amend the climate action plan of Ireland to require not just footprint reporting for large energy uses but also net accounting that looks at both the footprint and the positive handprint. This also extends into Ireland’s reporting to the United Nations Framework Convention on Climate Change, UNFCCC, so that, obviously, any claimed digital contribution is verified using a standard.
Fifth, it would be fantastic if Ireland could publish its own audited case study on the application of ITU-T standard L.1480 or the application of ITU-T standard L.1801 so the next time this question comes before the committee, Ireland has actually more evidence from its own use cases, not only from elsewhere.
Finally, it is important to define, measure and report on the twin transition language that is included in Ireland’s own strategies already in use. For example, Ireland’s digital AI strategy talks about the twin green and digital transition but never defines it, measures it or reports on it. The next revision to that strategy should really come up with a twin transition strategy with very clear metrics and a very clear reporting framework.
In conclusion, I believe the committee does not need to choose between the promise and the cost; it needs the accounting that tells the difference between the two and it is the accounting that helps Ireland to get the twin transition right. I thank members very much for this intervention. I am happy to take questions as part of the discussion.
