In our journey with ChatGPT for Confluence, we learned the importance of wisely choosing the data we fed into the system. Here are some key considerations and advice based on our experience:
Select Relevant and Current Data: When training the AI model, prioritize relevant and current documentation. Avoid feeding it with "work in progress" or obsolete information, which can lead to inaccurate responses and diminish the chatbot's effectiveness. Please review and update the data regularly to make sure it remains relevant and reflects the latest information within your organization.
Gradually Add Data: Rather than overwhelming the AI with a large volume of information at once, adopt a gradual approach to data feeding. Start with a manageable set of documents and incrementally add to it over time. This allows the AI to learn and adapt gradually, improving its accuracy and performance without being inundated with excessive data.
Incorporate External Documentation: Don't hesitate to leverage external documentation that is freely available and relevant to your organization's needs. Incorporating a diverse range of sources can enrich the AI's knowledge base and enhance its understanding of different topics. Please make sure the external documentation is free of copyright restrictions and compatible with the AI training process.
Regularly Evaluate Performance: Continuously monitor the performance of the AI model. Identify areas where the chatbot may be struggling or providing inaccurate responses and take proactive steps to address these issues. Regular evaluation and refinement are essential for ensuring the effectiveness of the AI-powered documentation system.
Encourage every relevant resource to send feedback: Reward collaboration and knowledge sharing within your organization. Empower everyone to contribute to the documentation process by suggesting improvements, flagging outdated information, and sharing relevant resources. By fostering a culture of collaboration, you can harness your team's collective intelligence to enhance the quality and accuracy of the documentation available to your AI.
With careful data selection, gradual refinement, and a collaborative approach, you can create a dynamic and accessible knowledge base that warrants your organization to thrive in the digital age.
Jacques Desormiere
Director of Quality Assurance