What the work involves
You will listen to and read AI-generated musical output — primarily Thai-language lyrics, sometimes paired with generated audio — and score it against a written rubric supplied by the AI lab. A typical task queue mixes several kinds of judgment: rating a lyric for creativity and craft, checking whether it actually followed the prompt (genre, mood, structure, subject), and flagging passages that appear to echo published Thai songs closely enough to raise originality concerns. A large share of the work is naturalness assessment: does this line sound like something a Thai songwriter would write, or like translated English? Are the slang, particles, regional expressions, and tone-syllable fit convincing when sung rather than read?
Expect to write short justifications for your ratings in English. Reviewers care less about your score than about whether your reasoning is specific and repeatable — "awkward register shift in line 3, formal ครับ against otherwise street-level phrasing" travels further than "sounds off."
What the platform screens for
- Verifiable music background. Songwriting credits, released work, performance or composition training, or a body of music journalism. Mercor's screen will ask you to name specifics and follow up on them.
- Genuine Thai fluency plus working English. The application includes a Bilingual Competency Interview conducted in Thai; instructions and rubrics are in English.
- Scene knowledge with range. Familiarity across contemporary Thai genres — luk thung, mor lam, Thai indie, T-pop, hip-hop, phleng phuea chiwit — rather than one narrow lane.
- Rubric discipline. Willingness to apply someone else's standard consistently, including when it conflicts with your own taste.
Logistics
Fully remote and asynchronous; you pick your hours. Most experts work around 20 hours per week and there is no stated cap. Start is immediate, duration up to six months, subject to project volume. You need headphones or studio monitors adequate for critical listening. Pay is observed at $18/hr and is not guaranteed — rates and task availability are set by the platform and the partner lab.