IntroductionAI tools are widely available. But without context, they produce patchy results. This post shares a simple three-part framework for using AI tools intentionally. It covers setup, prompting, and the most common pitfalls. It includes a real-world example, links to King’s institutional guidance, and a signpost to the DigiEd AI Literacy for All course for those who want to go deeper. AI Advantage SummitI recently attended the AI Advantage Summit hosted by Tony Robbins and Dean Graziosi. The summit was pitched at beginners and light to moderate users of AI, and provided several practical use cases from people who had used it to build businesses or run their personal and professional lives in a number of different contexts. Much of the content was useful both personally and professionally, but the most immediately applicable was a workshop run by Igor Pogany, in which he walked through how to set up an AI tool – specifically large language model chatbots like Copilot, ChatGPT and Claude – in order to get it to serve you better. This is Pogany’s framework in a nutshell. He provided a Google document into which you insert all the context for who you are, what you do, who you serve, what your goals are, and what you want out of the AI tool. This is cut and pasted into the profile text box in your chosen AI tool, which gives the tool the context it needs to serve you more effectively.  Then it’s time to get better at prompting. When starting a chat, make sure you articulate clearly:
We’ve all had experiences of using AI tools without this framework, and the results are patchy at best. Recently I showed a friend how to use an AI chatbot for the first time. A food writer and editor, she wanted to transcribe her handwritten recipes into a publishable book. We scanned in one recipe as an example. What it produced was unrecognisable: it changed the wording, reformatted the whole thing, and added erroneous photos from goodness knows where. This is an example of how the tool’s response itself shows you the gaps you haven’t filled in, both in the profile and in the prompting. In this case, what she hadn’t told it was what ‘good’ looks like. So we gave it a second try. She told it: I just want you to transcribe this. I am a writer and an editor — don’t mess with it, don’t add anything. You are my editorial assistant. You are not a creative force. This time, did what it was told, and then she was flying with it. The AI debate in the public sphere currently falls into two binary categories, and as technology enhanced learning professionals, we’re seeing this at the coal face. On the one hand are those, typically younger users, who seem happy to delegate everything indiscriminately to AI, with predictably mixed results. Which is why, on the other side of the argument, are those who are losing work to AI, or who see such indiscriminate use day to day and think: we’ve all seen Terminator 2. We know where this is going. The reality is that AI will never be, and was never designed to be, a substitute for human creativity. Like thinking itself, it’s a great servant but a lousy master. The backlash against AI is not unfounded, but it is unhelpful. In a recent meeting with other technology enhanced learning professionals I posed the question — not are you using AI, but how are you using it? I expected to find myself behind the curve. The reaction surprised me. Resistance. Suspicion, even. The resistance is understandable. But this is a movement that we as learning technologists should be engaging with thoughtfully, whatever our reservations. The OutcomePogany’s framework works well in both personal and professional contexts. At King’s, the official guidance for using AI tools is that you should be signed in with your KCL credentials, ‘improve for everyone’ should be switched off in privacy settings, and private institutional data should not be shared. Within those parameters, there are use cases worth gathering.  This post is one example. It came into being through a conversation about responsible AI use on a Sunday morning, which became a set of ideas, which I had summarised into bullet points, which became a dictated draft, which became the paragraphs you’re reading now. The scaffolding was provided by the AI. The thinking and the writing are human. This kind of collaborative, human-led process is a framework worth adopting in higher education and beyond. This approach aligns with King’s own stated commitment to the Russell Group’s principles on generative AI, which include equipping staff to support students in using these tools effectively, and – crucially – working collaboratively to share best practice as the technology evolves. The meso-level guidance goes further, actively encouraging staff to try these tools themselves, ideally in pairs or clusters, to understand their capabilities and vulnerabilities firsthand. This post is an attempt to do exactly that. For those who want to go deeper, the DigiEd AI Literacy for All course covers everything here, and much more. Module 2 focuses specifically on best practices for prompting. ConclusionThere is a productive institutional conversation to be had about responsible AI use in a professional context — one that moves beyond the binary and toward shared best practice. Our TEL network at King’s is well placed to lead it. Consider this post an opening contribution. What are your use cases? Useful LinksPlease note that the links below will open in a new browser window.
About the authorNaomi Grace Bignell is Learning Technologist for the AKC (Associateship of King’s College) programme at King’s College London. An artist, musician and writer, she is particularly interested in AI from a philosophical perspective.
|