How are you addressing prompt engineering in your organization? Is there a centralized prompt library or are team members learning skills to prompt on their own? Can you share any techniques or best practices your teams use for effective prompt engineering to maximize the output quality of GenAI tools?
Sort by:
We began by opening up tools like ChatGPT and Copilot to a broader audience, focusing heavily on upskilling in prompt engineering. We collaborated with vendors, including Microsoft, to provide training not just for IT engineers but for everyone. We emphasized that creating a good prompt isn't always necessary; you can interact with Copilot to refine prompts and reconstruct answers as needed. We chose not to develop a centralized prompt library due to the size of our organization and the diverse needs of our users. Instead, we focused on continuous upskilling through sessions on prompt engineering to empower our teams to leverage these tools effectively.
Our approach is quite similar to what Ramnath described. When we deployed Copilot, we noticed varied reactions. Some users loved it and used it daily, while others tried it once and decided it wasn't for them. Our current goal is to re-engage those who might have dismissed it initially by highlighting new features and teaching effective usage techniques. It's crucial to ensure readiness before a major push, as first impressions can be lasting, and we want to avoid losing users who might not return.
We don't maintain a prompt library but incorporate prompt engineering into our AI101 training, offering tips and tricks. We conducted a year-long pilot with a paid ChatGPT version, holding coffee sessions to answer questions and reinforce learning. We observed positive outcomes, with users realizing the tool's potential over time. We encourage continuous learning, suggesting LinkedIn courses on prompt engineering and AI. Change management and communication are vital in introducing tools like Copilot, as some people may be apprehensive or unsure where to begin. Our framework heavily emphasizes change management and learning.