Discussion

AI analysis of public comments finds data-centre opposition, but leaves a big question mark

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13822

An AI startup says it found widespread opposition to data centres by analysing public comments at government meetings, but its figures describe people who spoke at those meetings, not a representative survey of Americans. The findings offer a fresh way to map local debate, with a rather important asterisk beside the headline number.

Watch Desk analysis

What happened

KQED reports that San Francisco-based Prophecy Gov analysed more than 10,000 public comments from recordings of 906 government meetings across 417 entities. Its American Sentiment website assigns ratings to comments about data-centre projects. The company’s analysis found 90% opposition overall, while 79% of self-identified union workers supported development and 60% of self-identified data-centre workers opposed it.

Prophecy Gov did not respond to KQED’s requests for comment, and the website publishes no methodology. KQED says it is unclear how spoken comments were turned into a statistical rating. The analysis also omitted San Francisco and Oakland, despite public debate there.

Why it matters

The project applies AI analysis to public meeting records, potentially making a large and messy pile of local comments easier to explore. But public speakers are not a random sample of residents. As Jonathan Bash, chief executive of another AI company working with local governments, told KQED, meeting comments can show what engaged groups care about, but are less useful for predicting elections or reading residents’ minds.

That distinction matters when a striking percentage is presented as what “Americans” think. Oakland’s council unanimously extended its data-centre moratorium through 5 October 2027, KQED reports, showing that local concerns can have direct policy consequences even when this particular dataset cannot settle how widely they are shared.

Our read

This is a useful experiment in making local voices more searchable, not a national opinion poll. Treat the figures as a snapshot of recorded meeting comments until Prophecy Gov publishes its method, explains how comments were selected and rated, and shows how omissions were handled. “90% of speakers in this dataset” is less catchy than “90% of Americans”, but it is the sturdier sentence.

What to watch

  • Whether Prophecy Gov publishes its methodology and responds to KQED’s questions.
  • Whether the ratings can be checked against the original recordings and transcripts.
  • Whether the project adds omitted cities or other ways to hear from residents who did not speak at meetings.

Discussion spark: Should public-comment analysis be used to shape local data-centre policy before its methods are independently inspectable, or should councils treat it only as a map of who turned up to speak?

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.