Making the most of Generative AI

Getting Started with AI 

• When using Generative AI, search queries are known as 'prompts', and the writing of effective prompts is known as 'prompt engineering'.

• Prompts are the questions or instructions you input into an AI tool to generate text, images, code or any other type of content.

• Vague or poorly-structured prompts can result in the AIs generating incorrect or unhelpful responses.

• There are simple techniques you can use, such as the CLEAR model, which can help your write prompts to give you more useful and accurate responses.

Writing effective searches

Creating effective search terms and instructions is also known as 'prompt engineering' (Lo, 2023a). In other words, prompt engineering is the ability to generating search queries for AI systems like Chat-GPT or Bard. It is similar to searching in Google but requires a more refined approach. To construct effective and well thought out prompts they need to include several elements to produce meaningful and informative responses.

For example, if you were to ask ChatGPT for information about a specific topic, your prompt might be in the form of a question: "can you tell me about the history of artificial intelligence?" The model would then use the input prompt to generate a coherent and relevant response. It is important to note that the prompt is a crucial part of interacting with the model, and refining or providing more details in your prompt can often lead to better results in terms of the generated responses.

Using the CLEAR model

You can use Lo's (2023b) CLEAR model to help you create effective prompts:

Concise

A concise, focused prompt will remove all unnecessary information and allow the AI to concentrate on the most important aspects of the query and not to be distracted by irrelevant factors.

Basic prompt:

I am studying an introductory undergraduate module on economic theory. Please provide me with the basics of the capitalist theory aimed at a first-year university student.

CLEAR prompt:

Provide an introduction to capitalism for an undergraduate year one student studying economics.

Logical

A logical flow and order of ideas ensures that the AI system will correctly process the context and relationship between the concepts, resulting in more accurate and coherent responses.

Basic prompt:

Give me a recipe for a fruit cake.

CLEAR prompt:

Outline the steps involved in creating a recipe for a brandy-soaked fruit cake, starting with the ingredients and ending with a Christmas themed iced and decorated fruit cake.

Explicit

Give clear and explicit output requirements covering the format, content or scope.

Basic prompt:

Can you tell me about the first world war.

CLEAR prompt:

Write as if you are a GCSE History teacher. Produce a 50-minute lesson plan for a GCSE Year 11 class on World War One (1914-1918) from the point of view of Great Britain. Concentrate on the origins, key battles and aftermath of the war.

Adaptive

Give clear and overt prompts. Be flexible and experiment with key words and phrases. Be prepared to amend or build on previous prompts, depending on the results you get. 

Basic prompt:

Discuss the effects of climate change.

CLEAR prompt:

Discuss the effects of global warming in the last 100 years on the world's ocean levels, and its impact on marine life.

Reflective

Critically analyse the results you are getting from the prompt. Be prepared to evaluate the relevancy and applicability and rephrase a query if necessary.

Basic prompt:

Define bonding.

CLEAR prompt:

Define bonding in the context of chemistry.

References

Dianati, S & Luadari, S (7th Aug. 2023) An Introduction to prompting generative AI like ChatGPT for teaching and learning. THE Campus. https://www.timeshighereducation.com/campus/introduction-prompting-generative-ai-chatgpt-teaching-and-learning

Lo, L.S. (2023a) The Art and science of prompt engineering - a new literacy in the information age. Internet Reference Quarterly. https://doi.org/10.1080/10875301.2023.222761

Lo. L S (2023b) The Clear path: A Framework for enhancing information literacy through prompt engineering. Journal of Academic Librarianship, 49 (4). https://doi.org110.1016/j.scalib.2023.102720

McGowan, A; Gui, Y; Dobbs, M; Shuster, S; Cotter, M; Selloni, A; Goodman, M; Srivastava, A; Cecchi, G A and Corcoran, C M (2023) ChatGPT and Bard exhibit spontaneous citation fabrication during psychiatry literature search. Psychiatry Research, 326 https://doi.org/10.1016/j.psychres.2023.115334

Further Reading

Giray, L. (2023). Prompt Engineering with ChatGPT: A Guide for Academic Writers. Annals of Biomedical Engineering. 

A good article for those new to using AI, this gives an easy to read introduction to what prompt engineering is alongside a simple explanation of what large language models like ChatGPT are. It goes on to give examples of the different kinds of prompts and highlights some of the common pitfalls to avoid such as ambiguity, bias reinforcement, ethical issues and others.

Meskó, B. (2023). Prompt Engineering as an Important Emerging Skill for Medical Professionals: Tutorial. Journal of Medical Internet Research, 25(6).

his paper summarizes the current state of research about prompt engineering and, at the same time, aims at providing practical recommendations for the wide range of health care professionals to improve their interactions with LLMs. 

Dichone, P (2023) ChatGPT mastery: crafting effective prompts for optimal results [Video] Packt Publishing. 

This online course covers the fundamentals of ChatGPT, including creating an account and exploring the ChatGPT interface. It discusses the potential impact of AI on jobs and explores different prompt techniques such as specificity, instruction, contextual clarity, iteration, systematic prompt engineering, use of examples, and bias control.  The full course takes approximately 2 hours to complete, but you don’t have to work through the full course if you don’t want to. Particularly good are the videos on prompt techniques.