The Dynamics of Collective Creativity in Human-AI Hybrid Societies

TL;DR

This study investigates how human-AI interactions shape collective creativity in social networks by conducting large-scale experiments on $5\times5$ grids where stories are iteratively created and shared. Three conditions—human-only, AI-only, and human-AI—are evaluated using independent efficiency-rated creativity and semantic diversity analyses, including TF-IDF cosine similarity and UMAP embeddings. Findings show that AI-only networks have the highest initial creativity and diversity, but diversity in AI-only systems declines over time, whereas human-AI networks start lower but achieve the greatest final diversity, balancing continuity with novelty. The results demonstrate the potential of human-AI collaboration to enhance collective creativity and offer a scalable framework for studying hybrid cognitive ecosystems, while acknowledging ecological and methodological limitations that motivate future extensions to more persistent, multimodal tasks.

Abstract

Generative AI is shaping an increasingly hybrid society, where ideas and cultural artefacs are created both by humans and intelligent machines. Human creativity is influenced in complex, nonlinear ways by the actions of AI-driven agents within their social networks, but these influences are difficult to measure using traditional methods. This study examines how human-AI interactions shape the evolution of collective creation within large-scale social network experiments, where human and AI participants collectively create stories. Participants (either humans or AI) joined 5x5 grid-based networks in which stories were selected, modified, and shared over many iterations. Initially, AI-only networks showed greater creativity (rated by a separate group of human raters) and collective diversity of stories than human-only and human-AI networks. However, over time, hybrid human-AI networks became more diverse in their creations than AI-only networks. In part, this is because AI agents retained little from the original stories, while human-only networks preserved continuity. These findings highlight the value of experimental social networks in understanding human-AI hybrid societies.

Figures (4)

• Figure 1: Experimental framework for studying collective creativity. (A) Participants join social networks and engage in a creative writing task where short stories are selected, modified, and transmitted over many iterations (B) We study three network configurations: human-only, AI-only, and human-AI. (C) The creativity of stories is assessed by a separate group of human raters.
• Figure 2: The dynamics of collective creativity. (A) Mean creativity ratings of stories over time, as evaluated by human participants. (B) Diversity of stories (inverse similarity) over time. The horizontal axis represents the 25 iterations, grouped into five sets of five iterations each. Error bars represent one standard deviation, computed across participants. (C-D) Creativity and diversity gain: The improvement in measured creativity and diversity from the first iteration to the last.
• Figure 3: UMAP projection of the shared semantic embedding space, highlighting word clouds for specific clusters.
• Figure 4: Term dynamics by condition: Words are plotted along the horizontal axis, and generations are plotted along the vertical axis. A circle at position $(x, y)$ indicates that the word $x$ was used in a story at iteration $y$. A line denotes that the corresponding word was used by successive iterations. The size of the circle represents the frequency of the word’s appearance in the same iteration.