Consumer electronics giant trains content moderation AI with 30 moderators in 8 languages
A consumer electronics giant needed help to train its content moderation AI model to monitor its discussion boards and community more effectively.
The client’s discussion boards were riddled with Terms and Conditions (T&C) infractions, making it difficult to maintain and protect the company’s brand value without online community oversight and management.
The client required tech savvy content moderators located in-country and employed full-time to address discussion board issues and respond to customer service support inquiries – activities from which its content moderation model could learn. Moderators were required not only to remove policy-violating content, but also to communicate directly with community and discussion board users to provide basic customer service and support.
Challenges
- The client’s community and discussion boards were riddled with T&C violations, posing significant risks to the company’s brand
- Tech-savvy, in-country moderators employed full-time were needed
- Moderators were required to remove policy-violating content and respond to basic customer service support inquiries – activities from which the model could learn
Solution
- Human-in-the-loop validation
- TrainAI recruited, trained and now manage a global pool of tech-savvy, in-country moderators to:
- Review, edit, and remove unacceptable user-generated content
- Communicate with users who require basic customer service support
Results
- 30 full-time employed, in-country content moderators review:
- 500,000+ posts per year
- In 8 languages
- With 99% of posts processed within 5 minutes
- Well over 50,000 hours of work has already been delivered – work from which the client’s content moderation model continues to learn