Review and rate the actions and answers of a computer-using AI agent, verify claims by online research and write detailed annotations. Hourly long-term engagement, 16 weeks with possible extension; no rate published. 45 to 50 minute form assessment.
Hourly, 16-week engagement on a computer-using-agent evaluation project: review AI responses and actions, verify claims online and write ratings and explanations. A 45 to 50 minute Google Form assessment comes first. No rate published.
Korean-speaking photographers source licensed real-world images, obscure key details and write English prompts, rubrics and solution notes so multimodal agents must research to identify them. Contractor, 20+ hours a week, up to 16 weeks. No rate published.
Turing contract for illustrators, cartoonists and concept artists to make character artwork from reference images for an AI project. 20 to 40 hrs/week, paid in USD, portfolio review before onboarding. Beginners welcome.
Students and professionals with real Gemini learning history research, verify and correct AI responses. $30 per hour, 16 weeks, 20 hours a week (up to 40), paid in USD. Your Gemini conversation history is validated.
Evaluate Gemini's personalization in Portuguese using your own Google account data, ranking side-by-side responses and writing rationales. $15 per hour, 3 months, 4 to 40 hours a week with 4 hours of PST overlap.
Spanish speakers join scheduled, unscripted group conversations recorded for a multi-speaker diarization dataset. Turing flexible contract, remote, needs a wired headset or external mic. The ad publishes no pay rate.
French speakers take part in scheduled, unscripted group conversations that build a multi-speaker diarization dataset. Turing flexible contract, remote, headset or external mic required. No pay rate published.
Arabic speakers join scheduled, unscripted group conversations recorded for a multi-speaker diarization dataset. Turing flexible contract, remote, headset or external mic required. No pay rate published.
Japanese speakers join scheduled, unscripted group conversations recorded for a multi-speaker diarization dataset. Turing flexible contract, remote, headset or external mic required. No pay rate published.
Korean speakers join live, unscripted group conversations on assigned topics to produce multi-speaker diarization audio for AI voice training. Fixed session times and a strict microphone setup apply. The ad publishes no rate.
Chinese-language evaluation of Gemini's personalized responses: write prompts from your own context, rank pairs side by side and justify with turn references. $15 per hour, 3-month contract, up to 40 hours a week.
Turing contract for English and Bahasa Indonesian speakers to analyse content, verify claims online and correct language model answers. Contract of up to 12 months, US Pacific overlap, no pay figure published.
Turing contract for English and Tagalog speakers to analyse content in Tagalog, verify claims online and write corrections for LLM errors. US Pacific overlap, no contract length or pay figure published.
Turing contract for English and Russian speakers to analyse and summarise content, check claims online and correct language model answers. Contract of up to 12 months with US Pacific overlap; no pay figure.
Turing contract for English and Thai speakers who analyse and summarise content, validate claims online and write corrections where language models go wrong. US Pacific overlap; no contract length or pay figure published.
Turing contract for English and Malay speakers to analyse content, check claims online and give feedback that improves language model answers. US Pacific overlap; contract length and pay are not published.
Contract analyst role for bilingual Hebrew and English speakers: read and break down content, check claims online, and write correct answers with explanations that help train an AI language model. 40 hours a week, US-hours overlap. No rate published.
One-month contract for Finnish and English speakers who analyse content, check claims online and write corrected answers to improve an AI model. Up to 40 hours a week preferred, US-hours overlap. The ad lists its three-step screening. No rate published.
Contract role for Turkish and English speakers: summarise and fact-check content, build tricky analytical scenarios and write correct answers to improve an AI language model. Availability of up to 40 hours a week preferred, US-hours overlap. No rate published.
Remote contract role for Swedish and English speakers who summarise content, verify claims online and write corrected answers and explanations that make an AI language model better. Up to 40 hours a week preferred, US-hours overlap. No rate published.
Contract analyst work for Danish and English speakers: break down and fact-check content, then write correct answers and explanations to improve an AI language model. Up to 40 hours a week preferred, US-hours overlap. No rate published.
Contract role for Korean and English speakers: read and fact-check content, build scenarios models get wrong and write correct answers with explanations. Up to 40 hours a week preferred, US-hours overlap. No rate published.
Contract role for Japanese and English speakers: analyse and fact-check content and write corrected answers with explanations to improve an AI language model. Up to 40 hours a week preferred, 2 to 5 hours a day of overlap with Los Angeles time. No rate published.
New roles land most days. Get them on Telegram or Discord as they are added, or read how the platforms pay before you apply.