Ten Experiments with writing LLM Prompts for the Wolfram Prompt Repository

[kathryncramer]

I have been experimenting with writing prompts for the Wolfram Prompt Repository. I had two general goals:

• Creating assistants for cognitive tasks associated with strategy and design

• Creating prompts for conversational exploration

Here are some prompts for you to try. These are on submission to the Prompt Repository, although one or two under a different name:

Personas:

1. DesignThinker

2. InnovationConsultant

Frameworks

3. Causal Layered Analysis

4. Porter’s Four Corners

5. SWOT Analysis (this one doesn’t seem to work programmatically; I don’t know why)

6. Six Thinking Hats

Conversations
These can be used with Moodified or Setting

7. Simulate Conversation (flaky and may fail or crash Mathematica)

8. Panel Discussion (a collaboration with Bob Sandheinrich; this is flaky and may fail or crash Mathematica)

Modifiers

9. Setting

10. Topic

Here is a screenshot of an earlier version of prompts showing how the conversation and the modifiers can work together.

Suggestions welcome.

Bonus

Since apparently a lot of my prose made it into the training data I made a persona out of what GPT-4 says my writing is like. (The very flattering depiction of my writing style in the prompt definition is mostly what GPT-4 said. I added a few recent credentials to give it additional areas of expertise.) You can try out the Kathryn-bot if you like. (This one is not on submission.)

• Kathryn bot

Have fun!