thinking-hats-ai: Python package implementing six thinking hats prompting

Package

What is it?

thinking-hats-ai is a Python package that facilitates idea generation by following Edward de Bono's Six Thinking Hats methodology from his Book. It enables you to generate ideas by selecting one of the six hats and lets you choose one of the implemented prompting technique to follow while generating the idea.

Table of Contents

• Use of Package
  • Example script
  • Hats
  • Prompting techniques
  • Brainstorming-input
  • Developer mode
• Installation
• Dependencies
• Background
• Creators

Use of Package

Example script

This example uses the CHAIN_OF_THOUGHT prompting techniques and the BLACK hat for the personality. It also uses developer mode to log the interaction in a separate file.

### Import package
from thinking_hats_ai import BrainstormingSession, Technique, Hat, BrainstormingInput

### Create session
session = BrainstormingSession('YOUR-OPENAI-API-KEY')
session.dev = True # Activate dev mode

### Define current status
brainstormingInput = BrainstormingInput(
    question = 'How could you make students come to class more often even though there are podcasts provided for each lecture?',
    ideas=[
        "Implement an interactive class participation system with incentives",
        "Extra credits or digital badges, encouraging students to attend and engage actively",
        "Offer exclusive in-class activities or discussions that are not available in the podcasts",
        "Create a social media group for students to share their experiences and insights from attending class",
        "Organize regular contests or challenges related to class content, with prizes for participants",
        "Provide a comfortable and engaging classroom environment with refreshments and seating arrangements",
        "Incorporate gamification elements into the class structure, such as quizzes or team-based activities",
    ],
    response_length='5 bullet points'
)

### Generate output
response = session.generate_idea(
    Technique.CHAIN_OF_THOUGHT,
    Hat.BLACK,
    brainstormingInput
)

### Print output
print(response)

Hats

The different hats act as a predefined persona according to Edward de Bono's book about the six thinking hats in brainstorming. You can select which persona should be used for your instance.

source: Book

Prompting techniques

The different prompting techniques help to analyse different approaches of idea generation for each hat. While implementing, we analyzed which of the techniques work best for which hat.

Brainstorming-input

The instance of BrainstormingInput allows you to pass the brainstorming question, ideasand response_length to the generation of an idea.

Developer mode

The developer mode is used to log the in/outputs of the api calls. A log folder and log files will be created when executing a Script with activated developer mode. This was implemented for prompt engineering purposes and help to analyse the history of all API calls made.

It can be activated by setting the dev attribute to True (default False).

instance.dev = True

Installation

This package is available through the Python Package Index (PyPI).

pip install thinking-hats-ai

Dependencies

• LangChain - A framework for developing applications powered by language models
• AG2 - A multi-agent framework for building AI-driven systems
• Autogen - A framework for enabling autonomous AI agents to collaborate

Background

The implementation of thinking-hats-ai started at UZH as a part of three bachelors theses.

Creators

• Timon Derboven - timon.derboven@uzh.ch
• Leon Braga - leonroberto.braga@uzh.ch
• Marc Huber - marctimothy.huber@uzh.ch

Go to Top