Paul Murphy Questions AI Use in Electoral Boundary Reviews
Paul Murphy challenged the electoral commission's consideration of using AI to produce initial constituency boundary proposals, arguing it could create a privileged "first mover" advantage and affect democratic representation. He warned about limits to regulating AI and questioned using technology as a substitute for social resources.
Murphy said it is concerning to consider giving AI the role of drafting the first proposal for boundary reviews because an initial proposal has privileged status in subsequent debates. He argued that starting proposals shape committee reactions and amendments, and that privileging an AI-generated draft could influence outcomes.
The transcript records practical timing pressures: census results are due in May 2028 and the next scheduled general election is November 2029, while constituency reviews typically take around a year. The electoral commission noted the need to complete local electoral area boundaries, Dáil constituencies and European Parliament constituencies within a compressed period and said technology or algorithms might help speed the process.
Murphy echoed concerns that if engineers cannot fully explain AI decision-making, reliable regulation may be difficult, raising the question of whether unregulatable AI should be banned. He said he would not presume to legislate himself but framed AI as a diagnostic that raises broader legislative and ethical questions.
He challenged the idea of turning to AI because of resource constraints, asking why underlying shortages persist if technology is the proposed solution. Murphy suggested policymakers should examine whether lack of staff or funding — for example in education and teaching — is driving an over-reliance on "magic machines," and urged consideration of social resources that sustain contextual and relational approaches.
AI and first-mover advantage
Murphy said it is concerning to consider giving AI the role of drafting the first proposal for boundary reviews because an initial proposal has privileged status in subsequent debates. He argued that starting proposals shape committee reactions and amendments, and that privileging an AI-generated draft could influence outcomes.
Timing and procedural pressures
The transcript records practical timing pressures: census results are due in May 2028 and the next scheduled general election is November 2029, while constituency reviews typically take around a year. The electoral commission noted the need to complete local electoral area boundaries, Dáil constituencies and European Parliament constituencies within a compressed period and said technology or algorithms might help speed the process.
Regulation and possible bans
Murphy echoed concerns that if engineers cannot fully explain AI decision-making, reliable regulation may be difficult, raising the question of whether unregulatable AI should be banned. He said he would not presume to legislate himself but framed AI as a diagnostic that raises broader legislative and ethical questions.
Resources versus technological fixes
He challenged the idea of turning to AI because of resource constraints, asking why underlying shortages persist if technology is the proposed solution. Murphy suggested policymakers should examine whether lack of staff or funding — for example in education and teaching — is driving an over-reliance on "magic machines," and urged consideration of social resources that sustain contextual and relational approaches.
We publish thousands of recordings to make Irish politics transparent and resistant to manipulation. Spotted an error? Report it — together we are building a reliable archive of Irish politics.
Other speeches
Tego samego dnia All speeches from this day →
Carol Nolan
Carol Nolan: Calls Mercosur 'Poisonous Deal' Threatening Farmers
Danny Healy-Rae
Danny Healy-Rae Demands Equal €600 Support for Irish Families
Sharon Keogan
Sharon Keogan urges councils to build, criticises housing plan
Richard Boyd Barrett
Richard Boyd Barrett criticises housing crisis, corporate profits
Danny Healy-Rae
Danny Healy-Rae Vows to Fight Mercosur, Defends Farmers
Mattie McGrath
Mattie McGrath: Urgent Call to Cut NGOs Out of Farming Talks
Transcript
thanks a lot happy to get back for a second round um just just kind of related to that and to the electoral commission just you mentioned earlier about potentially using ai as the kind of first draft of boundary commission reviews because i accept there's a lot of data and it's a question of like numbers but also geography um but it is concerning to me um that you would consider that and because it's it's linked to what dr mccrillon is saying like yes i accept that that's only the first step there'll be a human that looks at it but obviously the first proposal has a privileged status um we all know that in all our work you can go to a committee and you bring the proposal people are shaping reacting to that proposal that's the starting point and then they can amend it this way or that way but that's the starting point and effectively you're saying you're at least considering i don't know if you've decided or not but you'd be open to the idea that we would give that kind of first mover advantage um to to ai of what is a very fundamental part of our whole democratic system the nature of our constituencies how tds are elected um so if you can expand a little bit on on that and if you'd share those concerns yeah um thanks there's um there's a question of time here you know they um the results of the census will be out in may 2028 and the next scheduled general election is november 2029 normally takes a year to do um the uh the constituency review there is huge and the the last couple have been done by crunching excel spreadsheets and maps etc etc we need to look at technology may not necessarily be ai i mean there there are algorithms etc um that might be able to help us speed up this particular process because after we report the oirectus has to legislate to introduce them and um the venice commission say that there should be no changes to constituencies for at least a year in advance of a general election so there's a particular challenge here which we do have to deal with you know so we haven't settled on what that might look like but we also have to do the local electoral area boundaries as well as the doll constituencies and the european parliament constituencies all within that very very compressed time period and if i can just offer you perhaps some reassurance i don't think there was a single constituency of the 43 where the first option lasted and sustained over the um the seven members of the commission looking at it i mean there are so many tweaks etc there is no benefit to first mover advantage in relation to that simple crunching of the numbers the first time out thanks i'm skeptical about that i think there's always a first mover advantage i think there is you you use the term phrase tweaking like you know i'm in the room you know and um i mean it was the first time we did it uh because the electoral commission is new and um i mean we look at issues we look at boundaries i mean things like country boundaries um were difficult for us and i mean the people of earlingford and freshford still keep me awake at night you know they're i hope the people of feathercarn do to uh do too um uh just back to miss dr dr mcquillen um you say even the engineers who build ai can't explain what's going on inside so reliable regulation is a non-starter and and that's i'm coming from a very similar place uh to that in terms of where i've conclusions i've come to so far anyway from our discussions um if it can't be regulated then should it be banned and how do you actually do that and what does that actually look like well i wouldn't presume to step on the toes of people in the room when it comes to legislating about things i think my my interest my emphasis really would be on uh taking ai as a diagnostic really and saying where people are and actually following on from our colleague from the electoral commission you know talking about situations where well you know ai does seem to be the an option here because after all we're under incredible pressure and we don't have enough resources and so on and so forth which is such a universal narrative wherever you're looking and certainly is in higher education and you know we've got whether this is ai thing we can turn to i'd say that's that's always a point at which people should be asking and this perhaps does rise up to the legislative level uh actually what else could we do you know if there really isn't enough time to do that why isn't there enough time to do that if there really aren't enough teachers why aren't there enough teachers you know is this really a matter of overall social resources i mean we're just talking about the amount of money that's flowing around the tech sector is it really that there's not enough resources to provide enough teachers or enough uh whatever it is that provides the contextual and relational way of doing things that historically and currently turns out to be far more reliable than turning to these magic machines thanks a lot thanks