AI Quality Analyst: English (Up to 40 Hours a Week)

What you'll do

You evaluate a personalization feature for Gemini, which tries to make answers more relevant by drawing on your past Gemini chats, Gmail, Google Search and YouTube activity. Your job is to design multi-turn starting prompts (typically 1 to 5 turns) from your own life, then judge whether the model used that context well.

Specifically, you check for grounding (are claims about you backed by evidence, or are they flawed inferences or hallucinations), integration (is personal data woven in naturally, or does the model "overnarrate" it in a robotic way), and helpfulness. You stack-rank two responses side by side, write rationales that point to specific turn numbers, extract and verify the model's "Debug Info", and delete your evaluation conversations afterwards so they do not pollute your future chat history.

Who qualifies

• Strong English reading and writing (English is the focus language)
• Willingness to use your primary personal Google account, not a test account, with personal data sources enabled
• Analytical judgment on nuanced, ambiguous answers; an eye for incorrect personalization, poor inferences and forced connections
• Clear, structured written rationales
• A BS/BA or equivalent experience in a relevant field (the ad suggests policy, law, ethics, linguistics, journalism, computer science or similar analytical fields)
• Experience in data annotation, AI quality evaluation or content moderation is strongly preferred
• A desktop or laptop with a good connection

What it pays

The ad does not publish a rate. Turing sets rates per project; see Turing.

Worth knowing

Good:

• Hours are flexible within a range: at least 4 hours a day, up to 40 hours a week
• The work is judgment and writing, not coding
• Remote, with a defined 3-month contractor engagement

Less good:

• The privacy trade-off is real: the project needs your genuine Gmail, Search and YouTube history, so think hard about what the model will be reading and what you are comfortable exposing
• The ad also says full-time availability in your local time zone is required, as Turing staffs a global 24-hour team, and it asks for 4 hours of overlap with PST; these sit awkwardly with the "from 4 hours a day" line, so confirm the real expectation
• No pay figure
• Contractor engagement, no benefits mentioned

The ad describes the process: shortlisted candidates receive a Job Interest Form, then an assessment to be completed within 24 hours, then a conversation about pre-onboarding requirements.

About this listing

Read from Turing's job board on 18 October 2026. Turing lists this role twice; this is the more recent posting, and the older one is published separately. The duties, hours and process are Turing's; the privacy comments are ours. See Turing.