What the work actually involves
You write and answer analytical problems that models currently get wrong, then explain the reasoning well enough that a model can learn from it. In practice that means three recurring task types: reading a long document or dataset and decomposing it into logical blocks; solving constraint and deduction puzzles (scheduling, ordering, allocation) and writing out the step-by-step derivation; and fact-checking claims through live online research, citing what you found. A second stream involves the Danish side — producing or reviewing the same class of reasoning in Danish, where idiom, number formatting, and phrasing conventions differ from a literal translation of the English.
Much of the output is critique rather than authorship. You will be shown a model response, asked to judge whether the answer is correct, whether the reasoning that reached it is sound, and whether a confident-sounding chain of steps conceals an unstated assumption. A right answer arrived at by broken reasoning is a failure in this work, and saying so clearly is the core skill.
What the screen is looking for
Turing's process leans on written and AI-led assessment before any human conversation. Expect an unaided reasoning task, a short written rationale, and a Danish-language sample. The platform screens for genuine bilingual fluency rather than schoolbook Danish, for arithmetic and data interpretation you can perform without a model doing it for you, and for feedback writing that is specific — pointing at the step where the reasoning broke, not registering a general impression. A bachelor's degree is preferred, not required; demonstrable professional writing or analysis history substitutes.
Logistics
- Contractor engagement, no paid or medical leave; extension depends on performance and project demand.
- Fully remote, asynchronous within the day, but 2–5 hours daily must overlap America/Los_Angeles (UTC-8).
- Up to 40 hours per week preferred; your own desktop or laptop and reliable connection are required.
- Pay is undisclosed on this listing. Rates on Turing's language-and-reasoning tracks vary by language and task tier and are agreed before onboarding — ask directly rather than assuming a band.