Hi there, folks. Welcome to my little presentation on prompt engineering. My name is Charlie Reynolds. I'm a fractional CTO. a big fan of AI, and very keen on prompt engineering. So let's dive into a little presentation just by way of introduction then. So just like in life, you get what you ask for. So if you ask good questions, you'll get good answers. And if you don't, then the saying follows. and for me, I think asking clear and imaginative questions leads to some interesting results and some useful results. And that unlocks a lot of good capability from AI. There's also, beyond that, you've also got AI orchestration, which you're going to touch on. And that's where you use multiple specialized AIs to work together. to do various jobs in concert. this is a quote that I quite like by Einstein. imagination is more important than knowledge. there's a little bit more context to this, but I think, the gist of it for what we're looking at right now is that by not just giving very specific instructions where you're looking for very specific results, but using a little bit of creativity can actually get you to some innovative outcomes. so with artificial intelligence, I think this is now particularly poignant. so the technical knowledge and skills are obviously crucial in order to build AI, and evolve it and so forth. But I think a little bit of human, imagination and ingenuity can go a long way. so when it comes to prompt engineering, asking the right questions in the right way. we'll get you to some very useful and valuable outcomes on, as I say, a bit of imagination goes a long way. in terms of what we mean by prompt engineering, in simple terms, it's about designing effective instructions that get you effective results. so asking the right questions, thereby unlocking value. and, it supports effective interaction. So in terms of some key concepts, as I say, it's not just about the technical skill set. It's also about some creativity as well as precision. and rather than just talking to AI about what we already know about, It's also asking ourselves, to use our imagination, look at what is possible. and that therefore means that by the two of you combining those skills of using human ingenuity with the vast amounts of knowledge that AI can tap into, you can get to some new possibilities and new innovations. So by way of examples, then, instead of asking AI very short, simple questions like what's the weather, which is obviously very vague, you should be much more specific and ask questions along the lines of, what's the weather in Paris next Tuesday morning for my business trip? and you'll get a vastly different response. quite often, there'll be a little, there can often be a bit of embellishment as well, where you'll find that, by adding in little details about the nature of what you're doing and why you're doing it and so forth. the AI can take those cues and actually give you more than you bargained for. it might not just tell you what the weather's doing, it might also tell you that, As it's a business trip in Paris, actually these are some good activities considering what the weather's likely to be doing. Or again, you can feed that in to refine your prompt. and again, that's just a very short, simple example, but you get into, you can get into much more complex examples. with the kind of work that I do, for example, I often find that using several different AIs together can be very effective. so you can get much more complex tasks. and that's where what's called AI orchestration comes into play, where you can find specialized AIs that you can string together. we'll dig into some examples of that in a moment. so things like a system that generates, text. So chat GPT is prime example of that. another that would analyze sentiment and then another that can actually summarize the results. By putting different prompts into each of those and taking the results from each of those and feeding them into the next AI, you can actually get to quite a powerful solution that way. How does that work then? How are we crafting effective prompts? there are many different ways to do this, and this is just my take on it. there are some, tried and tested methodologies out there and some, mnemonics that you all may be familiar with, but I'm coming at it from the angle of using your imagination and striving for innovation. There's three parts that I've used in this, which is clarity and imagination. clarity is key. Absolutely. But using imagination can get you a long way as well. so if you think of yourself like a director and you're guiding a bunch of actors. And you're envisioning and, imagining the final performance and then crafting prompts that lead the AI in that direction. so again, if we look at AI used for content generation, for example, instead of saying generate an article about AI, you might say something more write a 500 word article on AI's impact on healthcare using statistics from the last five years. And that is obviously much more specific. the next part is context and orchestration. as I say, in more complex tasks, you might be using multiple AIs that have each got their specialisms. and you have to have real clarity around what it is you're trying to do and exactly what it is that you're expecting from each of those AIs, for example, and therefore they will fit together neatly and give you what you need. as I much mentioned earlier, you can have, an AI that can provide summaries of articles so you can take vast amounts of information and knowledge, and turn those into succinct summaries. you could then use another AI that is more specialized around things like bias, and then you could pass that over to another AI that can actually pull all that together. So having taken a vast amount of knowledge. summarized it, checked it for bias, and then converted it into a final output, which is, free from bias. and you would use different prompts for each of those in order to get to that outcome. and then lastly, iteration and refinement. just as many of you are probably used to in the world of software development. iteration is key. and each time that you put that prompt through and refine it, then you'll get closer to the desired output. So again, an example around that could be that you're using a, chat bot in the customer service context. And again, you've got, chatbots that will happily interact with customers, but at the same time, you'll have a much more specialized AI that is finely tuned for sentiment analysis, and we'll be looking to see if that customer is getting frustrated. And therefore, in some cases, it might decide that actually this is where human intervention would be optimal. so you orchestrate those models together so that you have, much more, enjoyable experience for the customer and you have a much slicker AI solution. So why is prompt engineering important? why does it matter? Given that AI is going through this massive growth spurt, and it's now becoming quite ubiquitous. the ability to ask the right questions is becoming more and more important. So prompt engineering enables us to transform AI from a simple tool into a collaborator. and again, this is where I go back to the idea that Einstein put forward around imagination being so important. and, it's not just about what the A. I can do, but it's about what we can imagine it doing, in fields like health care, finance, customer service and many others. prompt engineering, as we've touched on briefly, allows us to orchestrate many different systems together with that AI capability, and therefore you get seamless and intelligent workflows off the back of that. in healthcare, for example, AI systems can be used to orchestrate patient data, predict outcomes and assist in diagnoses. and obviously the success of those systems depends on how well they're guided with the prompts, but also, it's critical that they've got high quality data. being fed into them as well. So moving on to conclusions. as I say, asking AI about things that we already know about isn't really getting the best value out of it. it's challenging yourself and challenging the AI to think differently. and to explore new possibilities, and asking the right question in the right way, and that therefore opens doors, orchestrates systems together harmoniously and leads to innovative solutions. what I would say is the next time you interact with AI, don't just ask it for an answer, imagine the possibilities. asking the right questions and thinking about how you might orchestrate different AI models together to create something greater than the sum of its parts. you could even ask AI to help you with refining those prompts, and say, how can I make this better? what's suboptimal about this? give me a crash course in how to refine my prompts. there are many possibilities, around how you can interact. With the AI, which will enhance your prompts, but it will also enrich the AI and it becomes a virtuous circle. So that's all for me. It's just a short, sharp taste of presentation. there are many different levels and aspects to this, but I just wanted to provide a bit of challenge around using your imagination. that's all for me, Charlie Reynolds of CTR Consulting. and I hope that I will speak to some of you in the near future. Thank you. Take care. Bye bye.