Getting the most out of an LLM
Lesson details
Learning outcome
I can describe how to improve LLM output and use prompt engineering to improve LLM output.
Key learning points
- The performance of LLMs is highly dependent on how users craft their prompts.
- Prompts should be specific and clear to enable the model to interpret what you want.
- Prompts should provide relevant context or examples so the model can generate more accurate and tailored responses.
- LLMs can reflect biases from their training data, so responses should be critically evaluated and cross-checked.
Keywords
Prompt - the input question, instruction, or message you give to a large language model (LLM)
Vague - when something is unclear or too general to be useful
Context - the extra information you give to help the LLM interpret your prompt better
Engineering - using knowledge and skills to design and create things that solve problems or make life easier
Common misconception
You can just type anything into an LLM and it will always give the best answer.
LLMs work best when you give them clear, specific and well-structured prompts. The quality of the input directly affects the quality of the output; this process of designing and refining inputs to get better results is called prompt engineering.
Teacher tip
Ask learners to use a vague prompt with a chatbot. Then get them to engineer and improve their prompt by adding specific details, relevant context and stating the format they would like the output to be in. Then compare the before and after outputs to see how their changes improved the response.
Equipment
Pupils would benefit from access to a device that can use an age-appropriate AI chatbot.
Licence
Lesson video
Loading...
Prior knowledge starter quiz
6 Questions
Q1.What do the initials LLM stand for in artificial intelligence?
Q2.Match the example to the correct keyword:
Q3.Which of these is an example of bias in an AI system?
Q4.What is the role of training data in an LLM?
Q5.Put these steps in order for how a chatbot generates a reply:
Q6.Which of these is NOT a reason to be cautious about trusting LLM outputs?
Assessment exit quiz
6 Questions
Q1.What is the most important factor in getting a useful response from an LLM?
Q2.What is the process called where you design and refine inputs to improve the responses from a large language model?
Q3.Match the keyword to its definition:
Q4.Put these steps in order for improving an LLM’s output:
Q5.Which statement about using LLMs is correct?
Q6.What should you do if an LLM’s answer seems biased or incorrect?
To help you plan your 10 computing lesson on: Getting the most out of an LLM, download all teaching resources for free and adapt to suit your pupils' needs...
To help you plan your 10 computing lesson on: Getting the most out of an LLM, download all teaching resources for free and adapt to suit your pupils' needs.
The starter quiz will activate and check your pupils' prior knowledge, with versions available both with and without answers in PDF format.
We use learning cycles to break down learning into key concepts or ideas linked to the learning outcome. Each learning cycle features explanations with checks for understanding and practice tasks with feedback. All of this is found in our slide decks, ready for you to download and edit. The practice tasks are also available as printable worksheets and some lessons have additional materials with extra material you might need for teaching the lesson.
The assessment exit quiz will test your pupils' understanding of the key learning points.
Our video is a tool for planning, showing how other teachers might teach the lesson, offering helpful tips, modelled explanations and inspiration for your own delivery in the classroom. Plus, you can set it as homework or revision for pupils and keep their learning on track by sharing an online pupil version of this lesson.
Explore more key stage 4 computing lessons from the Using data science and AI tools effectively and safely unit, dive into the full secondary computing curriculum, or learn more about lesson planning.