The audit eventually moved beyond technical teams when the city council scheduled a public hearing on CivicLearn.
Elias expected questions about model accuracy.
Instead, the first speaker asked why the city had built the system at all.
The question changed the frame.
Technical evaluation usually begins after a problem and solution have already been chosen.
Public scrutiny can reopen the prior question of whether automation is appropriate in the first place.
A teacher argued that the city had invested heavily in prediction while leaving counsellor vacancies unfilled.
If the model identified need more efficiently but support remained scarce, she asked, was the system solving the right bottleneck?
A parent asked whether the city was using automation to compensate for political unwillingness to fund universal support.
A student asked why an algorithm should decide who deserved attention before every student had access to basic resources.
Elias defended CivicLearn as a triage tool designed for limited capacity.
The word triage immediately provoked another objection.
Triage may be necessary under scarcity, but scarcity itself is often shaped by policy rather than nature.
Priya watched Elias realize that engineering vocabulary can quietly convert political choices into technical constraints.
Another speaker supported the system, explaining that his daughter had received tutoring after CivicLearn flagged difficulties her school had previously overlooked.
The hearing therefore did not divide neatly into supporters and opponents.
People distinguished among beneficial uses, harmful uses, missing safeguards, and broader institutional failures.
A disability advocate asked whether students could request alternative review if the model relied on behaviour affected by disability.
A civil-liberties lawyer asked what legal basis authorized the collection and combination of data across agencies.
A school administrator warned that eliminating the tool entirely might return the city to opaque human decisions with even less oversight.
The council asked the audit team to explain not only performance metrics but purpose, authority, alternatives, and proportionality.
Elias presented fewer graphs than he had prepared.
He spent more time describing assumptions, limitations, and appeal rights.
The hearing converted system design into democratic contestation, forcing technical assumptions to encounter values that no optimization function could settle.
Scarcity, efficiency, and triage ceased to be background conditions and became political claims requiring justification in public rather than silent acceptance in code.
The hearing taught him that public accountability begins where technical competence ends: at the question of who gets to decide what the system is for.