Generative AI (GenAI) tools are not the same as search engines. Through using natural language processing (NLP), GenAI tools use Large Language Models (LLMs), to convert the prompts (instructions) you give them into useful outputs. Some tips:
The ICES outline for prompts can help give better quality outputs:
I: Identity/Audience |
Give the bot a role to play, and a target audience/level for output. |
C: Clear Commands |
Give clear criteria or outcomes for the work, using command terms. |
E: Explain & Examples |
Explain clearly what you want to achieve. Include quotes, details, examples, context. |
S: State Structure |
Define the format of the response (bullets, table, paragraphs etc). |
Each GenAI tool will behave differently, so choose tools according to your task:
Remember: the more context you give the bot, the more useful the results will be.
Prompts for Vocab Lists
Prompts for Bilingual Vocab Lists
Summarize or Rewrite Text
You can also use Diffit for this type of task.
"Read" or Summarize Links
This is not always reliable - it depends on the quality of the source and the model you use. Treat with caution.
Ideation & Brainstorming
Use the ICES structure to generate prompts.
Example for Perplexity or BingChat: (example result here)
[I] I am a [level, subject] teacher developing a course to connect [topics]. [C] Develop a course outline that includes [essential elements], with a suggested resource list that links to reliable resources. [E] The course needs to address the following objectives [paste objectives]. It also needs to address the following key and related concepts [paste concepts and their definitions]. [S] Present your results in the form of an introductory overview and a table that outlines how the topic can be addressed through the key and related concepts. Provide a reference list.
Finding video resources for class in Perplexity:
Use the ICES prompt and give clear guidance to adapt this: Suggest videos of [X-Y Mins] suitable for [age level] to....
Suggesting Activities
Suggesting Connections
Checking for Comprehension
Generating Exemplars & Mentor Texts
Set up the ICES prompt for context, and adapt with some of these ideas:
Example texts might be used in class to help students tune-in to the assessment and recognise the descriptors in the work. You could use a different topic from the classroom assignment.
Remember: the more context you give the bot, the more useful the results will be.
Suggesting Connections
Checking for Comprehension
Considering Multiple Perspectives
Set up the ICES prompt, and try the following adaptations:
Suggesting Interesting Outputs
Suggesting Imagery
Ideation & Brainstorming
AI Tools should NOT be used for final assessment. Our students deserve the attention and care of their teachers. However, students and teacher might use AI tools for ongoing feedback, conferencing, support, with permission. Consider:
Giving quick feedback on writing
Giving feedback for improvement based on descriptors
This one is complex, but can be useful.
I am a [level] student preparing a [assignment type] for [class]. Give feedback on my draft writing to help me improve. Identify strengths and areas for improvement, with suggestions, agains the following objectives:
Objective 1: [paste descriptor]
Objective 2: [paste descriptor]
Present the results as a table, with rows for each descriptor and columns for strengths, areas for improvement and suggestions. Here is my draft: [paste text].
Remember: the more context you give the bot, the more useful the results will be.
Prompts for Bilingual Vocab Lists
Adjusting Reading Levels
Supporting Multilingual Learners
Identifying Key Terms
Explain at 4 Levels
This strategy can support understanding of complex topics at different levels of complexity. Adapted from Wired Magazine's 5 Levels series.
Remember: the more context you give the bot, the more useful the results will be.
This tab will be updated with advanced prompts for use with ChatGPT Plus and Advanced Data Analysis. An example is here.
A collection of bots, made in Poe, to support teaching and learning.
Try Poe here: https://poe.com/
In early GenAI, "Prompt Engineering" was touted as a desired skill. This might not be true for much longer, as LLM models adapt and become more capable at interpreting the user's needs. Through chatbot interfaces, users can iterate and give feedback, to get closer to their desired outcomes. Multiliteracies, global knowledge, strong writing & communication skills and the ability to learn and iterate might be more useful attributes.