Our Approach to AI

Where it's used, why, and how you always know which is which.

AI is automation, extrapolated

We consider AI to be one utility of many. In fashioning tools and especially in the age of computing, we have always had AI/automation/programming on hand, in some way, shape, or form — recommendation engines, auto-tagging, spellcheck, mastering assistance. What's changed recently is the scale and visibility of it, not whether it exists. We think that's worth saying plainly instead of pretending the line between "automated" and "not automated" is cleaner than it is.

Where it's already part of this site

Not hidden, not a future roadmap item — already running, in tools you can use right now:

  • The catalog tool uses AI matching to find your releases across platforms and pull titles, URLs, and ISRCs automatically — you confirm each match, it doesn't just assume.
  • The Spotify Exporter and related tools lean on automated lookups the same way.
  • Some curator dispatches are tagged AI notes — AI-assisted listening passes, clearly labeled as such, sitting alongside human-written picks.

Every pick, labeled by source

This is the actual position, not "AI good" or "AI bad": you should always be able to tell which is which, without having to guess. Every dispatch on the site carries a source tag:

Human criticA person listened to this and wrote the note themselves.
AI listening notesAI-assisted — the note reflects an automated listening pass.
Community pullSurfaced from what listeners are actually saving and sharing.

None of these is treated as more legitimate than the others by default — they're just different, and worth knowing apart. A good AI-assisted note and a good human note are both good notes. A bad one of either kind is still bad. Labeling doesn't rank them; it just tells you what you're looking at.

Why this, instead of a blanket policy

A lot of companies are putting out AI statements right now, and most of them land as either "we don't use it" (often not fully true) or "we use it responsibly" (often not specific about how). We'd rather just show the actual mechanism — the tags are right there on every dispatch, not explained once on this page and then never checked again.

If something's mislabeled, or you think a tag should say something different, tell us — we're happy to help.