SIP Study Group - 11th September 2025
Meeting assets for SIP Study Group - 11th September 2025
Quick recap
Winton, a cybersecurity professional and educator, introduced a session on ChatGPT focusing on prompt engineering and provided an overview of his background and the group's mission. He led a detailed session on effective AI prompting, covering topics like prompt quality, specificity, and the importance of clear instructions for obtaining accurate AI responses. Winton demonstrated various practical applications of AI prompting, including email rewrites, lesson plans, meal plans, and task management, while emphasizing the importance of maintaining quality standards and refining prompts incrementally.
Next steps
- Attendees to practice creating prompts using the CRISP+E framework for their specific use cases.
- Attendees to build their own templates for recurring tasks and save them to their prompt library.
- Attendees to complete the hands-on exercise of rewriting "Help me eat healthier" using the CRISP+E framework.
- Attendees to review the quality of their AI outputs using the 1-5 scale rubric presented.
- Attendees to book a 15-minute discovery call with Winton if interested in further discussion about career goals.
Summary
ChatGPT Prompt Engineering Insights
Winton, a cybersecurity professional and educator, introduced a session on ChatGPT, focusing on perfecting prompts for generative AI models, particularly GPT-4. He emphasized the importance of iterative refining processes and shared his experience in prompt engineering. Winton also provided an overview of his background, including his roles as a program director, senior auditor, and founder of Cyberspoke, and highlighted the group's mission to assist individuals in advancing their cybersecurity careers. He encouraged participants to book a 15-minute discovery call to discuss their goals and explore the community's resources.
Mastering AI Prompt Engineering
Winton led a session on effective AI prompting, emphasizing the importance of well-crafted prompts in obtaining accurate and reliable AI responses. He explained the difference between weak, better, and great prompts, highlighting how small changes in phrasing can significantly impact the quality of AI-generated answers. Winton also discussed the concept of the prompting spectrum and shared examples to illustrate the importance of clear and specific instructions for AI. The session included live demos, hands-on practice, and a rubric for self-evaluating prompts, as well as a Q&A session to address commonly asked questions.
AI Instructions for Effective Outputs
Winton discussed the importance of providing specific and detailed instructions when using AI models, particularly for tasks like creating a presentation outline. He emphasized the need to define the context, role, and structure of the output, as well as any parameters or constraints. Winton also stressed the importance of evaluating the AI's response for accuracy and quality, and advised double-checking the results to ensure they meet the desired specifications.
Kyoto Trip Planning Strategy
Winton discussed the importance of planning a trip to Kyoto with explicit criteria such as budget, balance, and feasibility. He emphasized the need for context, including factors like season, weather, and travel companions, to create a realistic itinerary. Winton also highlighted the importance of defining inputs, outputs, and constraints for any task, and suggested omitting unnecessary information to focus on what matters. He provided an example of preparing a 1-page summary for the Executive team, using confirmed figures from the Q3 financial report and limiting it to 300 words.
Optimizing AI Task Efficiency
Winton discussed how to effectively use AI by shaping its expertise and experience level, emphasizing the importance of specificity and alignment with audience needs. He advised keeping tasks singular and explicit to reduce errors, using delimiters for task sequences, and structuring outputs for clarity. Winton also highlighted the importance of considering cultural differences when using AI and suggested requesting alternatives for outputs to visualize options.
Optimizing AI Task Prompts
Winton provided guidance on crafting effective prompts for AI tasks, emphasizing the importance of clarity, simplicity, and avoiding anti-patterns like vague instructions or multiple goals. He suggested keeping prompts concise, treating AI as a resourceful intern, and using source citations to prevent hallucinations. Winton also introduced the "crispy framework" and quality bar concepts to help define success criteria and input materials for AI tasks. The discussion concluded with a demonstration on email rewrites, highlighting the benefits of using prompt libraries and saving successful prompts for future use.
ChatGPT for Executive Email Rewrite
Winton discussed using ChatGPT to rewrite an email for an executive audience, focusing on maintaining a professional tone and avoiding blame. He highlighted the importance of reviewing the AI's rationale for the rewrite and ensuring clarity in next steps. Winton also emphasized the need to double-check the revised email's length and content to maintain quality and alignment, particularly when involving other parties.
AI-Generated Lesson and Marketing Content
Winton demonstrated how to use AI to generate lesson plans and marketing copy, emphasizing the importance of clear, specific prompts and double-checking outputs to ensure they meet expectations. He showed examples of generating a comprehensive lesson plan on photosynthesis and marketing copy for a product, highlighting the AI's ability to provide structured outputs with suggested materials and resources. Winton also explained how to refine prompts to make them more thorough or lengthy if needed, stressing the value of starting with a clear foundation for the AI to build upon.
Personalized Meal Plan Template Demo
Winton demonstrated how to create a personalized meal plan using a template, emphasizing the importance of specifying dietary preferences, restrictions, and context such as budget, location, and lifestyle. He shared his experience of customizing a plan to include more beef, chicken, and vegetables while avoiding cottage cheese and incorporating protein bars. Winton also highlighted the need for realistic meal plans that account for eating out and adjusting to changes in income or preferences.
Optimizing ChatGPT for Task Management
Winton demonstrated how to use ChatGPT for task management and template creation, explaining that users can build their own templates and save them in a prompt library for repeated use. He emphasized the importance of maintaining a quality bar, suggesting a target score of 4 out of 5 for outputs, and advised participants to avoid sharing sensitive information with AI models. Winton also recommended starting new chats for each project and refining prompts incrementally when needed. He concluded by announcing that the next session would focus on applied prompting and creating daily tools with AI templates.
0 comments