Large language models like ChatGPT, Copilot and Google Gemini are powerful tools, but how we use them significantly impacts the quality of the response we receive. Here's why going beyond simple transactions and fostering a conversation unlocks their true potential:
1. Context is King:
A single question lacks the richness of a conversation.
By engaging in a back-and-forth, you provide context that allows the AI to tailor its responses. They can consider past exchanges, understand nuances, and ultimately deliver something more relevant and insightful.
2. Unveiling the Underlying Need: Sometimes, what you directly ask for isn't what you truly need. A conversation allows the AI to delve deeper and identify the underlying goal of our requests. This means they are more likely to provide a more comprehensive solution that addresses the root cause.
3. The Power of Refinement: Conversational interactions enable us to refine.
As the AI provides feedback or asks clarifying questions, you can adjust your prompt, leading to a more precise and effective outcome.
4. Building on Existing Knowledge:
A conversation is like stacking building blocks.
Each question or statement adds to a foundation of knowledge that the AI can use to inform subsequent responses. This ongoing dialogue means the AI can be more helpful and relevant as the interaction progresses.This is the prompt framework I show teachers all over the world that takes their AI interactions from transactional to conversations.
First of all, let’s go on a 3 step journey to get to that framework.
Step 1 - The Free Flow Prompt
This is where everyone starts.
A free flow prompt is when you ask an AI chatbot to do something for you and it’s just an outpouring of unfiltered thoughts. Such as:
I need a lesson plan for a 1 hour Geography lesson on the 2008 Sichuan earthquake for a group of year 8 students.
You soon realise that the response to this will be a vague plan with little to no consideration for pedagogy, relevant content or the needs of your students. And let’s be realistic, how could it be anything more? It can’t read our minds.
*For tasks a lot more simple e.g. writing a generic letter, or for starting a conversation with AI, these types of prompts aren’t all bad.
Step 2 - The Transactional Framework
Through trial and error or a professional development session you might learn to structure prompts a bit more and learn to include certain elements in them.The framework I developed in December 2022 was the PREP framework.
I am very proud of this framework as it has helped thousands of educators around the world write better prompts and therefore transform their workload.
The framework, which I included in the best selling The AI Classroom, was one of the first of its kind but it was never meant to be comprehensive. How could it be when talking to these AI tools is similar to coaching someone for having a conversation with a colleague. Using a framework in these scenarios is always superficial. However, they can be useful for helping dive deeper in our pursuit of more impactful interactions. Diving deeper will always involve conversations.
Step 3 - The Conversational Framework
Based on this concept and backed by a research paper from November 2023 I extended PREP to PREPARE.
The framework still included the transactional PREP to start the conversation off, but then shifts to a dynamic conversation.
The ‘Ask’ and ‘Rate’ parts help to turn the agency in the conversation fluid. It is passed back and forth between you and the AI as you both reflect and evaluate responses, digging deeper each time.
Finally, an appeal to emotions on average gets a 10% better response. A quirk of an AI model that wants to please and will attempt to meet the stakes when you raise them.
My one hour workshop on the PREPARE model have been very popular.
I’d love to see you at one soon.
For now, I go more in depth in this video: