One of the central challenges in the study of co-creative AI is deceptively simple: How do we measure creativity that emerges through interaction?
Traditional approaches to evaluating creativity often focus on finished artifacts such as paintings, musical compositions, designs, stories, or generated outputs. However, co-creative systems fundamentally change the nature of creativity itself.
In co-creation, creativity does not arise solely from isolated individuals or autonomous systems. Instead, it emerges dynamically through interaction between coupled participants. This means that the most important aspect of co-creation is often not the final artifact alone, but the interaction dynamics that produced it.
How collaborators coordinate, improvise, adapt, influence one another, and construct meaning together through time becomes central to understanding creativity itself. This realization led to the development of Creative Sense-Making and the broader effort to quantify co-creation in situ.
Before the emergence of co-creative AI research, most computational creativity systems were evaluated similarly to autonomous generators. Researchers typically examined output novelty, artifact quality, surprise, value, or human judgments of creativity. These approaches worked reasonably well for systems generating artifacts independently.
However, co-creative systems introduced a fundamentally different situation.
In co-creative interaction, humans adapt to AI, AI adapts to humans, interaction histories matter, meaning evolves dynamically, and creative trajectories emerge recursively through participation. The resulting creativity belongs neither solely to the human nor solely to the machine. Instead, creativity emerges through interaction itself.
This created a major gap in the literature: there were few frameworks capable of modeling and quantifying the dynamics of open-ended creative collaboration.
In 2017, Nicholas Davis and collaborators introduced Creative Sense-Making: Quantifying Interaction Dynamics in Co-Creation at the ACM SIGCHI Conference on Creativity and Cognition.
The paper proposed one of the earliest formal frameworks specifically designed to quantify co-creative interaction through time.
The framework synthesized enactive cognition, participatory sense-making, improvisational collaboration, distributed creativity, and interaction analysis into a new cognitive framework for modeling co-creation dynamically.
Rather than treating collaboration as a static event, Creative Sense-Making approached co-creation as a continuously evolving process of mutual adaptation, coordination, interpretation, and meaning construction.
The framework focused specifically on interaction rhythms, styles of turn-taking, coupling dynamics, cognitive states, and the emergence of shared meaning through time. This represented a major shift in how co-creative systems could be studied.
One of the most important conceptual contributions of this work was the idea that co-creation produces activity traces.
Every interaction during collaboration leaves behind a measurable trajectory of participation. These trajectories can include drawing actions, timing patterns, interaction rhythms, feedback exchanges, conceptual shifts, and evolving coordination structures.
Over time, these traces form creative trajectories.
Creative trajectories describe the evolving path of co-creative interaction as collaborators adapt to one another and reorganize meaning dynamically through participation.
This idea helped establish a new subfield within co-creative AI research focused on modeling interaction histories, quantifying collaborative dynamics, tracking creative evolution through time, and visualizing the emergence of co-creation itself.
Rather than analyzing creativity solely as static output, researchers increasingly began studying the temporal structure of collaboration. This interaction-centered approach significantly expanded how co-creative systems could be evaluated and understood.
Perhaps the most significant methodological contribution of Creative Sense-Making was the development of sense-making curves.
Sense-making curves provided a way to model and visualize the cognitive dynamics of collaboration continuously through time.
The framework drew heavily from enactive cognition, participatory sense-making, and the free-energy principle to conceptualize how cognitive agents dynamically fluctuate between different interaction states during co-creation.
The curves tracked interaction dynamics such as cognitive coupling, improvisational exploration, interaction stability, engagement, and adaptive coordination. This allowed researchers to analyze co-creation not as isolated events, but as continuously evolving trajectories of participation.
The significance of this contribution was substantial.
Prior to this work, many evaluations of co-creative systems relied primarily on surveys, interviews, artifact judgments, or static outcome measures. Sense-making curves enabled researchers to quantify interaction dynamics in situ, observe fluctuations during collaboration itself, and analyze how meaning emerged dynamically during interaction.
This created a fundamentally new methodology for studying co-creation.
One of the most important aspects of the framework was that quantification occurred during interaction itself.
Rather than evaluating creativity only after collaboration had ended, Creative Sense-Making analyzed the unfolding dynamics of participation continuously through time.
This enabled researchers to ask entirely new kinds of questions:
When do collaborators become strongly coupled? How do interaction rhythms influence creativity? What patterns precede conceptual breakthroughs? How do humans respond differently to co-creative agents? What interaction structures increase engagement and emergence?
The framework therefore shifted evaluation away from static outcomes toward dynamic interaction processes. This distinction became foundational to later work in co-creative AI.
The Drawing Apprentice became one of the earliest systems through which these ideas were operationalized.
The system was not only a co-creative drawing agent. It also became a research platform for modeling and quantifying co-creation itself.
The Drawing Apprentice captured interaction histories, turn-taking structures, conceptual shifts, collaborative adaptation, and evolving activity traces during drawing sessions.
This enabled empirical investigations into participatory sense-making, emergent meaning, cognitive coupling, and collaborative trajectories.
Subsequent research expanded these ideas through conceptual shift modeling, explainable co-creative systems, interaction trace analysis, and quantified collaboration frameworks.
Together, these works helped establish quantified co-creation as an emerging research direction within co-creative AI.
The broader significance of quantifying co-creation extends beyond computational creativity alone.
The framework introduced a fundamentally different way of thinking about intelligence, collaboration, and creativity.
Traditional computational paradigms often assume that intelligence resides inside isolated agents, creativity is internally generated, and outputs are primary.
Creative Sense-Making instead proposed that intelligence emerges through interaction, creativity unfolds through participation, and meaning arises dynamically through coupled activity.
This interaction-centered perspective increasingly aligns with modern developments in human-centered AI, hybrid intelligence, participatory AI, interactive machine learning, and embodied cognition.
As AI systems become more adaptive and collaborative, the ability to model interaction dynamics may become increasingly important.
The early work on Creative Sense-Making helped establish many ideas that continue shaping co-creative AI research today, including activity traces, creative trajectories, interaction-centered evaluation, participatory sense-making, quantified collaboration, and dynamic co-creation modeling.
What began as an effort to understand collaborative drawing and pretend play evolved into a broader framework for understanding how creativity emerges through interaction itself.
The core insight remains profoundly important:
Creativity is not merely a property of isolated humans or machines.
It is a dynamic process that unfolds through participation between them.