Nicholas F.
Exploitative workflow disguised as opportunity
I joined OpenTrain AI for their Mandarin Transcription Project. The onboarding looked legitimate — funded milestone, clear instructions, and an apparent link to reputable clients. However, once inside, the structure became clear: • Training and onboarding were unpaid, despite taking hours. • Pay was conditional on 100% acceptance of each microtask, with no pay for corrections or partially correct work. • The “expected time per task” (5–10 minutes) was unrealistic, especially given the dense tagging rules and quality expectations. • Communication was vague, and deadlines kept being extended while new workers were continuously added, creating the illusion of progress but effectively crowdsourcing free labor. I passed their quality test with a perfect score, yet it was obvious the system was designed to favor quantity over genuine linguistic skill. The data contributors are treated as interchangeable, even though their work trains AI speech models. In short: If you value your time or your professional expertise, think twice before joining.
Response from OpenTrain AINov 18, 2025
Hi Nicholas, -Payment for fixing tasks: If tasks were incorrect and you were asked to fix them, all contributors were paid for that time -The expected time per task across the whole project was reasonable, as the final time per task across the project was 6.5 minutes