Discussion

Tucson school-closure consultant used ChatGPT to analyse public feedback

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13481

A consultant hired by Tucson Unified School District used ChatGPT to help analyse community feedback on proposed school closures, according to board member Sadie Shaw and the consultant’s account to the Arizona Daily Star. The practical question is whether the summaries captured the full range of what families said, not whether AI appeared in the process at all.

Watch Desk analysis

What happened

The district hired Bill de la Cruz for $75,000 to advise it and collect feedback on potential closures. He led four community-engagement sessions in September, three in person and one online. The district has proposed closing seven schools, with a vote scheduled for 8 December.

De la Cruz told the Star that AI and other electronic tools helped manage written feedback, identify recurring topics, compare comments across sessions and produce preliminary thematic summaries. He said those tools supported the work but did not replace professional judgement. Board president Ravi Shah said ChatGPT was used to synthesise feedback from hundreds of participants. Superintendent Gabriel Trujillo described it as an assistive tool for efficiency.

Shaw said she was concerned about using ChatGPT and questioned whether the analysis might have left out negative feedback. At an October meeting, board member Jennifer Eckstrom asked to see all the information gathered, rather than only positive feedback. De la Cruz’s contract includes further work after the closure vote, contingent on the vote taking place.

Why it matters

This is a concrete case of generative AI being used to summarise public input on a consequential local decision. The reported use was to organise and synthesise feedback, not to decide which schools should close. But a summary can still shape what decision-makers notice, especially if board members cannot inspect the underlying comments or see how the themes were selected.

The consultant and district leaders describe AI as support for a human-led process; board members have raised questions about whether the resulting account is complete. Those are different claims, and the available report does not establish that feedback was omitted. It does show why transparency about source material and review matters when AI helps turn many voices into a shorter list of themes.

Our read

Using AI to sort hundreds of comments may save time. It does not make the summary self-explanatory, or remove the need to check what it leaves out. For a decision affecting seven schools, the useful standard is straightforward: show the themes, make the underlying feedback available to the board, and explain how human reviewers checked the summary. A neat synthesis is only useful if it has not tidied away the disagreement.

What to watch

  • Whether the board receives the full set of community feedback and further detail on how it was analysed.
  • How the 8 December closure vote proceeds and whether board members address concerns about the summaries.
  • What further work the consultant carries out if the vote allows the next phase of the contract.

Discussion spark: When AI helps summarise public feedback, should decision-makers be required to publish the underlying comments and explain how they checked the summary?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.