Our Human Creativity Is Becoming More Uniform Due To ChatGPT
Our ideas, solutions and artistic expressions are becoming less original & diverse.
Introduction
One of the primary use-cases for ChatGPT is to use it to become more creative, or to generate new and unique ideas.
This recent study considers how, instead of ChatGPT making us more creative, it leads to similar ideas across disparate users. It also leads us to approach and experience the creative process differently.
In a study with 36 participants, the researchers found that users of ChatGPT produced less semantically distinct ideas compared to alternative creativity support tools (CSTs).
Additionally, ChatGPT users generated more detailed ideas but felt less responsible for them.
The Problem
The challenge is that a large number of people are using highly centralised, data-driven AI systems (such as ChatGPT) for our creative ideas and content. This leads to decreased diversity in the results of our creative processes, amongst other things.
Below on the right, is a representation of users making use of ChatGPT to produce much more homogenous ideas. At a group level users on the left are making use of more traditional creativity support tools with more diverse ideas.
Key findings are:
The use of ChatGPT as a creativity support tool (CST) might make the LLM’s users feel more creative.
Making use of ChatGPT can even broaden the range of ideas suggested by each individual user.
But also homogenise ideas and impede our natural creative journey.
Algorithmic Mono-Culture
Concerns about LLM-driven homogenisation of creative outputs have been discussed in terms of algorithmic monoculture.
Using a single AI system like ChatGPT for tasks traditionally performed by diverse humans can lead to more uniform outcomes. This effect might also apply to creative processes if large numbers of people use the same LLM-based creativity support tool (CST).
Beyond monoculture, LLM-based CSTs may cause homogenisation due to several factors…
The creative design processes often experience fixation effects, where initial ideas limit variation in later solutions. LLMs can induce fixation by presenting complete-seeming ideas early, reducing diversity in subsequent ideas.
LLMs may also diminish valuable ambiguity in human-machine co-creation by producing text that appears “finished.”
Trust in LLMs, due to their authoritative style, might lead people to accept LLM-suggested ideas without critical examination.
Moreover, groups are less creative than individuals when they converge on majority opinions, and LLMs trained to produce likely results might reduce creativity by appearing authoritative.
In Conclusion
The study shows that LLM-based CSTs strongly homogenise the ideation process.
This homogenisation also happens at the group level, possibly due to the similarity between LLM outputs and finished ideas.
AI technologies can aid creative processes, but improvements in CST design and model development are needed to fully support creativity.
Seemingly there exists an opportunity for an AI-based creativity support tool which does not present the user with the complete creative solution. But rather the envisioned tool should fuel the human creative process or arc by leading the human through a process.