Use the CLEAR-principles for effective prompts
Use the CLEAR-principles for effective prompts
On this page we introduce a prompting framework consisting of five principles known as the CLEAR-principles. The purpose is to improve your interaction with the chatbot.
Read more about AI literacy under the heading "prompting and iteration". (LINK)
- 1
Be Concise
Leave out unnecessary information to let the AI model focus on the core issue. An unclear input with half-baked ideas will give you a muddled output.
- Remember: Define exactly what it is you want an answers to in a clear academic language without any superfluous details.
- 2
Be Logical
Your input must have a logical flow. A logical structure enables the model to understand the context and the mutual relations between the concepts you are working on.
- Example: Describe the order of a process step by step, as opposed to naming the elements in random order.
- 3
Be Explicit
If you want to make a long prompt, you should state your output expectations clearly including scope, format, and theoretical perspective. It reduces the risk of receiving unexpected, irrelevant, and generic answers.
- Example: Rather than writing "Tell me about renewable energy", you should write "Identify five sources of renewable energy and make a brief description of how each of them work, in table format".
- 4
Be Adaptive
Interaction with AI is an iterative process. Do not accept a superficial answer. Instead, use the answer to adjust your next question, make it more focused, or ask the model to apply a different perspective.
- Remember: The model remembers previous imputs from the current conversation. If you want to change the subject, you should start a new chat to avoid irrelevant noise from earlier prompts.
- 5
Be Reflective
In terms of your own learning, this is the most important step. What sounds good today, may not sound as good tomorrow, because the technology changes all the time. Keep analysing the answers you receive: what type of wording results in useful answers, and when did the model misunderstand your prompt?
Linguistic devices: Boost your dialogue with the model
Apart from the CLEAR-principles, there are a number of linguistic devices that you can apply to boost the model's perspective and format. These devices are particularly useful when you use AI as an academic sparring partner.
Boost perspective by assigning roles
You can influence the focus of the model by assigning explicit roles; either to yourself or to the model.
Define your own role: For example, if you write "I'm a sociology student writing a paper on... ", you provide the model with a framework regarding your academic context, so it can adjust the response accordingly.
Assign an academic stand to the AI: Ask the model to apply a specific perspective, e.g. "Act as a critical peer-reviewer and find weaknesses in my argument."
Important ethical boundaries: You can ask the model to apply the perspective of a specific theorist (e.g. "Provide an analysis based on Foucault's concept of power"), but you cannot use the AI to generate fictitious opinions and apply them to actual persons mentioned by name.
Test comprehension with analogies and target groups
Asking the model to explain concepts and texts in alternative ways is quite efficient if you are faced with difficult material.
Change the target group: For example, you can ask the model to: "Explain quantum physics as if I were a high school student." This can ease your way towards grasping difficult theoretical concepts. .
Use analogies: Use the AI to "translate" abstract concepts into something more familiar: For example "Use a biological analogy to explain the operations of a database."