AI Quality Analyst: English

What you'll do

The project tests how well Gemini personalizes answers using what it knows about you from past Gemini conversations, Gmail, Google Search and YouTube. You write starting prompts from your own experiences, usually running 1 to 5 turns, and then act as the critic of the model's reply.

Three things are under the microscope: grounding (claims about you must be supported, not hallucinated or inferred badly), integration (personal details should feel natural, not narrated back at you), and overall helpfulness. Much of the day is side-by-side comparison: pick which of two responses is more helpful, easy to use and enjoyable, and justify it in writing with turn-specific references. You also pull the model's "Debug Info" to confirm chat summaries and data sources were really used, and delete your test chats so they do not skew your later history.

Who fits

• Fluent written English, since English is the project language
• Ready to use your main personal Google account, not a throwaway, with personal data sources switched on
• Comfortable judging subtle, ambiguous output, including forced connections and poor inferences
• Meticulous enough to spot small differences in naturalness
• A degree or equivalent experience in a relevant analytical field; prior annotation, AI quality or moderation work is strongly preferred

What it pays

No rate is given in the ad. Turing sets rates per project; see Turing.

Worth knowing

Good:

• Unusual task: your own life is the test material, so no specialist subject knowledge is needed
• Flexible hours within 4 per day and up to 40 per week
• Defined 3-month contractor engagement, remote

Less good:

• Your personal email, search and viewing history sit inside the evaluation, which is a privacy decision to make deliberately
• The ad asks for full-time availability in your local time zone (a 24-hour global team) and 4 hours of overlap with PST, which is not the same as "a few hours a day"
• No pay figure
• No benefits as a contractor

Turing's ad outlines the process: a Job Interest Form for shortlisted candidates, a 24-hour assessment after profile review, then pre-onboarding discussion.

About this listing

Read from Turing's job board on 18 October 2026. This role has been open since February 2026, and Turing also lists a newer posting of the same role, which we publish separately. Duties and process are Turing's; commentary is ours. See Turing.