This is a focused companion
This page is a focused companion, not the main overview of annotator work. Main data-annotator guide. It covers one practical decision: which questions to clarify in writing when you look at two different kinds of listing that the portal groups under the annotator family.
The descriptions below were opened on 2026-10-01. Historical listing descriptions do not confirm current jobs, pay, eligibility or interview steps. Treat them as a snapshot of what those listings asked on that date.
The two activities the sources describe
Contributing your own media. One set of listings asks participants to take and submit photographs of themselves following simple project instructions, then complete a short questionnaire about themselves and the submitted images, using their own smartphone (listing 2; also 3, 4, 5). Your own body and face become the subject of the dataset.
Annotating media produced for the project. Another set asks contributors to review and label short video or audio clips, and to assess AI-generated descriptions (listings 8, 9, 11). A separate group works on footage of robotic arms performing tasks, applying photography or filmmaking principles to judge image quality and document visual nuances (listings 1, 13). The source material here is produced for the project, not by you.
What each activity asks you to clarify
The two activities differ in what “clear instructions” means, and therefore in what you should ask.
For the own-media activity, the questions centre on the subject and on who controls it:
- What exactly do “simple project instructions” look like — poses, settings, lighting, number of images?
- Which questionnaire fields are required, and can they be answered partially?
- How is a task judged “approved”? Is approval appearance-based, technical (focus or exposure), or consent-based?
- Who owns the submitted images, and what uses does consent cover — including any use to test or train systems, or to build visual profiles?
- How is eligibility applied in practice — region, age band, or smartphone model?
For the annotated-media activity, the questions centre on the rules applied to each clip:
- Which grading guidelines or scoring guide govern my assignment, and where do I read them?
- What documentation format and protocol must each annotation follow?
- How are ambiguous or uncertain clips handled — escalate, record uncertainty, or guess?
- What counts as “inappropriate” material I should flag, and who decides?
- If I correct AI output, what should the correction look like — full rewrites, targeted edits, or just a verdict?
Draft questions to clarify
The examples below are illustrative exercises created for this guide, not published micro1 assignments or interview questions. They are ungraded drafts of questions you might send — not required company wording, and not tests that a written example can evaluate listening accuracy or overall job readiness.
Draft 1 — own-media, image instructions:
“Before I start, could you share the written photo instructions for this project? I want to confirm the required number of images, allowed backgrounds, and whether I may use portrait mode.”
Draft 2 — own-media, consent:
“I’m willing to submit images and complete the questionnaire. Could you confirm in writing who retains usage rights to the photos, whether they will be used to test AI systems, and whether I can remove or withdraw them later?”
Draft 3 — annotated-media, uncertain clips:
“For the video clips, what should I do when a clip is too blurry or too short to label confidently? Should I mark it uncertain and continue, or report it instead of labeling?”
Preparation checklist
Keep this separate from what the listings require — it is editorial preparation advice, not company wording.
- Save a copy of any instructions, guidelines, or scoring guide you are given, with the date you received it.
- Note the device and software you will use, and check whether that satisfies the listing’s stated requirements (a smartphone for the own-media activity).
- Draft your consent and availability answers once, in writing, so you can reuse them consistently.
- Keep sample annotations or questionnaire responses that reflect the format you agreed to, before you start.
- Decide in advance how you will handle an uncertain item, and make sure your recorded handling matches what you agreed to.
Payment
The two activities this guide compares are represented by listing 2 (own-media) and listings 8-11 (annotated media). Listing 2 states a compensation unit of one approved task, while listings 8, 9 and 11 do not state a paid unit. No listing among those selected states an advertised hourly range, a weekly minimum, required availability, or the acceptance criteria used to judge a task — that is unknown rather than empty, and worth asking about. A different snapshot listing (12) does state a contract pay-per-task model; listing 5 states a per-participant per-task rate for a different dataset, but neither of those terms transfers to the two activities here.
Practical written questions that change this section: what is the paid unit for this listing; who accepts the work and by what criteria; what happens to work that is returned or rejected; and what minimum, if any, applies.
Effective time and effective hourly rate are different from the contract price per task. If a task takes longer than you expected — because the guidelines are stricter than they looked, or because you paused to clarify something — the price per accepted task is unchanged, but your effective earnings per hour fall. Extra time per accepted task can affect effective earnings per hour, not the contract price per task. Never conflate those units.
Ask micro1 for the payment conditions in writing before accepting work.
Next step
Deciding which listing to pursue is a separate question from deciding which questions to raise. For a fuller picture of what annotator work involves, read the Main data-annotator guide.