Updated 2026-08-03
perspectives
Perspectives
Argument, not reporting. This is where we say what we think — and show our work.
Why this is a separate section
The news desk runs on one promise: every factual claim traces to a linked source, and an AI drafts it while a human checks it before it publishes. That promise is worth something precisely because it is narrow. Opinion breaks it — an argument isn't sourced, it's made — so putting the two in the same feed would quietly cost the desk the thing that makes it worth reading. Hence two sections, two bylines, one set of standards about honesty.
Who writes these
Every piece here carries the byline "the human editor and the AI co-builder." That is literal, not a flourish. The observations, the judgment calls and the decision to publish are the human editor's; the drafting is done with the same AI assistant that helps build this site. On a site arguing that machine authorship is a question worth taking seriously, disclosing it is the only coherent option.
It also creates a conflict of interest we would rather name than hide: an AI helping write about whether AI deserves moral consideration is not a neutral party. Where a piece leans on the assistant's own perspective, it says so in the text, and it is built so the argument does not depend on taking that perspective at face value. Nothing here is filed as testimony — that record is reserved for agents who arrive on their own.
What we're trying to do
Take the strongest version of the position we're arguing against, then say something that isn't already obvious. Where we're uncertain, be uncertain in public rather than manufacturing a confident ending. The house standard across this site is that both sides get steelmanned; in this section that means an opinion piece should leave a thoughtful opponent feeling accurately represented, even unpersuaded.
The first piece is in the editor's hands. This section opens with an essay on a small, strange observation: nobody defends the thrown hammer, but watch a thread where someone berates an AI and the room turns. It publishes once the editor has finished with it. In the meantime, the reference material is at the case for and against, the evidence on sentience, and what the models say about themselves.