What the work actually involves

You write questions at the edge of your own discipline — the kind that require methodological nuance, multi-source synthesis, or reasoning a model cannot pattern-match its way through. Each item ships with a verified answer, primary-source citations, and an explicit reasoning trail that a reviewer who is not in your subfield can follow and defend. You then run your own question against current AI systems: if a model answers it cleanly, the item is too easy and you raise the difficulty without loosening the accuracy of the answer key. That test-and-escalate loop, not the initial drafting, is where most of the hours go.

Reviewer feedback is part of the cycle. Items come back for ambiguity in phrasing, citations that lean on secondary summaries, or answers that are technically correct but admit a second defensible reading. Revising cleanly and quickly matters more than being right the first time.

What micro1 screens for

The screen is AI-led and pushes on depth. Expect to name your subfield precisely, describe your research output, and then be asked follow-ups that a generalist could not answer — a contested finding in your area, why a common method fails under a specific condition, where the primary literature disagrees. Screeners also probe evaluation judgment: can you tell a hard question from a merely obscure one, and can you distinguish a model's genuine reasoning failure from a badly worded prompt. No prior AI evaluation experience is required and saying so costs you nothing; overclaiming it does.

Logistics

  • Fully remote, asynchronous, contractor engagement.
  • Output-based pay: per accepted task, not per hour. The $40–90/hr band is observed effective rate and varies with how fast you work and how often items pass review on first submission.
  • A weekly minimum submission volume applies.
  • Hiring moves fast — roles typically fill within 48 hours, with first tasks expected 24–48 hours after onboarding.
  • Written English at publication standard is non-negotiable, since ambiguity in phrasing is the most common rejection reason.