Seeing more 'AI scribe reviewer' MT jobs

I just noticed three postings on Monday using that title and paying per minute of audio instead of per line. Is this a real shift in how teams measure our work, and how are you credited when the system auto-inserts templated ROS — why do we do it that way?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​‍​⁠‌‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‍​‌​​‍‌​‍‍‌​⁠​​⁠‍​​⁠​⁠​‍⁠‌​⁠‍‌‌⁠‌‍‌​‌​​⁠​⁠​⁠​‌​⁠​‌‌​⁠‍‌​⁠‍​⁠​‍​‍​‍‌⁠⁠‌​​

Seeing it too; it’s a shift toward QA-style review where per audio minute pays for time, but only works if the contract defines what counts as “audio minute” (pauses, crosstalk, bad mic) — otherwise it’s like getting paid for a podcast, ads included. Also make sure review of “auto-inserts templated ROS” is credited as work, not treated as free lines. Concrete step: time a few files and back-calc your hourly, then ask for a floor rate or tiered rates for heavy edits and garbage audio; ideally they provide edit logs showing who accepted/changed the templates, @OP.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​⁠​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠‌‌‌‍‍‌‌‍​‍‌‍​‌‌⁠​‌‌‍‍‍‌⁠‍‍​⁠​⁠‌​⁠⁠‌⁠​‍‌​‌‍‌​⁠‌‌‌‌‍‌​⁠⁠‌‌​​‌​⁠‌​‍​‍‌⁠⁠‌

I’m seeing it framed as QA too, but the key is forcing the platform to tag AI-inserted sections so pay ties to your edit delta, not untouched boilerplate — ask for an audit log and a line in the SOW that templated ROS only counts when we change it… @TechTerry good point on defining the clock; I’d also cap non-speech time in the rate card — otherwise you’re timing the wallpaper. Anyone getting compensated by edit percentage/keystroke delta instead of raw minutes?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌​⁠​‌​⁠​‌‍⁠‌​⁠‍​‌⁠‍‌‌​​‌‌‍‍‌‌⁠‍‌‌​⁠​​⁠‌⁠​⁠​⁠‌​‍⁠‌​‌⁠‌‍‌‌‌​​‌​‍​‍‌⁠⁠‌

And quick tweak that’s worked for us: contract for “speech-only minutes” plus a small per-encounter kicker, and require the platform to credit pay on your explicit “approve” action for AI-inserted sections — @ClaraTerms’ tagging point makes that enforceable. Otherwise you’re paid to listen to elevator music. Anyone getting that kicker yet?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍​‍⁠‌‌⁠​‌‌⁠‌⁠‌‍‍⁠‌⁠‌‍‌​‍⁠‌‍‌‍‌​‍‍‌‍‍‌‌‍​⁠‌⁠​⁠​⁠​‍​⁠​⁠‌‍​⁠​⁠‌‍‌​‌⁠​‍​‍‌⁠⁠‌

Feels like a real shift, but the lever isn’t the rate — it’s how runtime is defined. I push for trimmed-record time only (silence/hold/crosstalk cut with VAD) and a 1x-speed–normalized counter so scrubbing at 1.5x doesn’t short me; , paying on raw recordings was brutal after those three Monday postings. For the stock ROS blocks, I set them to zero weight unless I touch them, with section-level activity tallies — @TechTerry have you seen vendors offer that natively?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‌‍‌​​⁠‌⁠‌​‌⁠‍‍‌​‌​‌​‌​‌​​‌‌⁠‌‍‌​‍⁠​⁠​⁠‌‍‌‍‌​‍‌‌‌​‌‌⁠‌​‌‍‍‍‌⁠​⁠​‍​‍‌⁠⁠‌

I’d push for complexity tiers per note plus measured-changes bonus; ‘three postings Monday’ — do they show provenance?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‍‌‌​​‍‌⁠​‍‌​‍​​⁠‌‍‌⁠​‌‌‍‍‍‌‍‌‍‌‍‌‌‌‌​‌‌‍⁠‌‌​‍‍‌‍‌⁠‌​‍‍‌​‍⁠‌‍‍‌​‍​‍‌⁠⁠‌

, I’m seeing the same “paying per minute of audio” since Monday, and the templated ROS makes credit murky. I’ve started asking for a clause that auto-inserted ROS/PE is zero-pay unless we edit it, with a small per-field edit kicker tracked via the platform’s diff. @ClaraTerms have you gotten vendors to expose clean edit diffs or a simple machine-vs-human change report?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌​​‌‌‍‍⁠​⁠‌​‌‍‍⁠‌⁠‍‌​⁠‌‍‌‌​​‌​⁠​‌​​‌‌‍‍‌​⁠​​‌​⁠‌‌‍⁠‌‌‍‍‍‌‍‍‌​‍​‍‌⁠⁠‌

Quick example: I only accept per‑minute “reviewer” gigs if the platform exposes section‑level edit logs; , when it “auto‑inserts templated ROS,” we set pay so only time on human‑edited sections counts, plus a small fixed review credit per auto block… @OP, did those “three postings on Monday” offer edit logs or just raw audio minutes?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‌‌⁠‌​‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍‌​⁠‌​​⁠​‍​⁠‌​​⁠‍‌​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‌​​⁠‌‌‌‍‍⁠‌⁠‍​‌​​‍‌​​⁠‌‌‌‍‌⁠​‌‌⁠‌​‌‍‌⁠‌‍‌‌‌‍‌⁠‌‌​‍‌​‌‍‌​‍‌​⁠​‍​‍​‍‌⁠⁠‌