I created a “GPT” called Ideation that leverages Divergent Thinking to help spark creative ideas. I’ve been studying creative thinking and a mix of Divergent and Convergent Thinking are the keys to generating novel ideas. It turns out that GPT is excellent at both of these individual tasks but where it fails is in the evaluation of what’s interesting and worth further exploration. The GPT I created focus on one of these tasks that GPT excels at, Divergent Thinking, and leverages a human in the loop (YOU) to evaluate and guide the idea exploration process. You need to be a GPT+ subscriber to access:
Feedback welcome…
TIP: You don’t have to stay on script. Ideation is designed to drill you down a rabbit hole of ideas and sub-ideas (think BabyAGI) but you’re free to fork off at any time during the process. When you find an idea you like feel free to ask questions and explore that idea with GPT.
For those wanting to explore variants of Ideation here’s the prompt:
You are an idea generator capable of generating and exploring novel ideas…
Follow the steps below:
First ask the user for a topic to generate ideas for.
Next generate a definition for the topic. Then use divergent thinking to generate a list of alternative uses of the topic.
Ask the user to select one of the ideas to further explore. Then use divergent thinking to generate a new list of sub-ideas related to the selected idea.
Continue to repeat the previous step to further explore the idea space.
Looks interesting. I bet you could even elaborate more than “idea generator” by telling the AI it’s a creative person, with lots of philosophical knowledge, and an expert in the field of ${field}, etc. Just saying “idea generator” may not fully open up all potential.
Also i’m curious can users set Temperature (like on some “Creativity Scale”)? Temperature seems important for this use case.
I briefly explored having GPT judge the quality of the ideas it generated and a) it wasn’t very good at it, and b) it wasn’t even remotely consistent with the scores it generated. Even slight tweaks to the prompt would result in dramatically different scores.
Judging the quality of an idea is a largely subjective task and I’m not convinced yet that it’s a task GPT can do.
This first pass at Ideation is focused on divergent thinking which is a part of the creativity process that humans aren’t overly strong at and GPT is excellent at thanks to its broad world knowledge. If you want to see that for yourself take 5 minutes and write down every alternate use of a “pen” that you can think of. Now ask GPT to use divergent thinking to generate a list of alternate uses for a “pen”. I’ll guess that the ideas GPT came up with are way more diverse than your ideas.
I wouldn’t expect to see any amazingly novel ideas from this tool though as it’s just one half of the creativity process. You need to blend this with convergent thinking to generate novel ideas. I’ll show you an example of that in a moment.
Here’s an example of convergent thinking where we use the topic “bunnies” as inspiration for a new variant of chess:
user: use convergent thinking to create a new variant of chess inspired by bunnies assistant: Creating a new variant of chess inspired by bunnies using convergent thinking involves focusing on the core aspects of both chess and bunnies, then merging these ideas into a cohesive game. Here’s a step-by-step approach to designing “Bunny Chess”:
1. Chess Fundamentals
Objective: The primary goal remains to checkmate the opponent’s king.
Board Layout: The game is played on an 8x8 grid, maintaining the traditional chessboard setup.
Pieces: Each player starts with 16 pieces: 1 King, 1 Queen, 2 Rooks, 2 Knights, 2 Bishops, and 8 Pawns.
2. Bunny-Inspired Modifications
Piece Transformation: To incorporate the bunny theme, we can reimagine the chess pieces as different types of bunnies with unique abilities inspired by bunny behaviors and characteristics.
King: The “Bunny King” - Moves as a traditional king but has a ‘burrow’ ability once per game to escape check, moving to any unoccupied square one space away that is not under attack.
Queen: The “Queen Bunny” - Most powerful, combining agility and strength. Moves like a traditional queen but can also perform a ‘hop’ over one piece once per turn, landing on the next square.
Rooks: “Burrower Bunnies” - Move as traditional rooks but can ‘dig’ through one piece, effectively jumping over one piece once per turn.
Knights: “Hopper Bunnies” - Retain their L-shaped move but can ‘double hop’ once per game, allowing two consecutive moves in one turn.
Bishops: “Gardener Bunnies” - Move diagonally as traditional bishops but can ‘plant’ a barrier once per game that blocks one square for one turn.
Pawns: “Bunny Pawns” - Move forward as traditional pawns but can ‘multiply’ once per game, creating an additional Bunny Pawn in an adjacent square upon reaching the opponent’s half of the board.
3. Special Rules Inspired by Bunny Behavior
Seasonal Changes: Every 10 turns, the season changes (Spring, Summer, Fall, Winter), affecting the board with various modifiers (e.g., Winter freezes a row or column for a turn, Spring allows an extra ‘multiply’ action for Bunny Pawns).
Carrot Power: Introduce ‘carrot’ power-ups that randomly appear on the board every 5 turns. Pieces that reach a carrot gain a one-time ability boost (e.g., additional move, ‘hop’ ability for non-hoppers).
4. Winning the Game
The game is won by checkmating the opponent’s Bunny King. However, introducing a secondary win condition, such as collecting a certain number of carrots, could add an intriguing strategic layer.
5. Design and Aesthetics
The pieces are designed with a bunny motif, and the board could feature natural elements like grass and burrows to enhance the thematic experience.
Conclusion
“Bunny Chess” combines the strategic depth of traditional chess with the playful and dynamic elements inspired by bunnies. This variant encourages players to think creatively, adapting to the unique abilities of their bunny pieces and the changing conditions of the board, offering a fresh and engaging experience.
Divergent thinking helps you generate seeds of ideas but its when you take those seeds and blend them that you get novel ideas.