Removing the Auto-training feature
In our ongoing efforts to ensure the highest quality and accuracy of your chatbot's responses, we've decided to deprecate the Auto-training feature.
A few key factors influenced this decision:
- Precision and Control: Auto-training, at times, added less relevant training questions to Answers, which could impact the bot's response precision. We believe in providing clarity and direct control over the chatbot training process, ensuring richer interactions.
- NLU Advancements: Our newly improved Natural Language Understanding (NLU) model now boosts accuracy in predicting declarative Answers by 5-8%, eliminating reliance on automatic training. The enhanced comprehension makes Ada's predictions even more accurate.
What this change means:
- Ada bots will no longer self-train based on chatters' thumbs-up feedback.
- On the Ada dashboard, under Settings > Bot setup > Answers, we've removed the Auto-training section.
When training your bot, we recommend doing the following:
- Train your Answers with 15-20 sentences that capture the range of ways customers reach out for support.
- Regularly review and refine your Answer training. For a step-by-step guide, please watch our video walkthrough: Fine-tune your bot's training.
For more information on how to effectively train your bot, see Training best practices.