Representation & Equity
More people are starting to reach government through an AI agent, whether to take part in a consultation or to claim a benefit, license, or appeal, because they can and because they expect it to carry their point further. As that grows, an agency has to respond, adapting how it gathers requests and how it reads them. What an agent does to a voice is not settled: a structured, polished submission might read as clearer and better argued, or arrive stripped of the markers that made it distinctively theirs. The harder question is who can adapt to whichever way that falls.
A citizen with no agent, a weak one, or one built on a model that represents them poorly has the least room to adjust, and is often the citizen a service most needs to reach. Left alone, this could pull a service toward the voices that are best equipped, and away from the people it exists to serve.
Government owes people roughly equal standing to reach an outcome it controls, whether to be heard in a consultation or to transact for a benefit, license, payment, or appeal, an equality that has rested on citizens reaching the service directly.
As more participation comes to run through a privately supplied agent layer that government neither distributes nor controls, whether a citizen's request carries will increasingly depend on whether they have a capable agent at all, on how well a model represents people like them, and on whose interests the agent serves in carrying their position, none of which the service itself sets.
An agency that kept handling requests as they arrive would let standing in its own services track the quality of a citizen's tooling, and could no longer meet its duty to treat the public on equal terms without acting on the agent layer directly.
Make sure every citizen can reach a government outcome, whether to be heard in a consultation or to transact for what they are entitled to, whether they bring a capable agent or none at all.
Keep what the citizen said intact regardless of the intermediary.
Make any agent route usable whatever a citizen's language, literacy, or assistive-technology needs.
Guarantee a route without an agent that reaches the same outcome, not just the same access.
Account for who a model serves badly, and keep those citizens capably represented.