The publications below by Nicholas Davis, PhD and colleagues trace the development of co-creative AI, Creative Sense-Making, enactive AI, participatory interaction, quantified collaboration, and human-AI co-creation across more than a decade of interdisciplinary research conducted by Nicholas Davis and collaborators spanning computational creativity, cognitive science, human-computer interaction, and human-centered AI.
These works document the emergence of an interaction-centered paradigm for artificial intelligence — one in which creativity and intelligence arise through dynamic participation between humans and AI systems rather than isolated generation alone. Developed through collaborations with researchers across computational creativity, design, cognitive science, and AI, this body of work helped establish many of the foundational concepts now associated with co-creative AI and enactive approaches to human-AI interaction.
Many of these works were developed in collaboration with researchers at the Georgia Institute of Technology, the University of North Caroloina at Charlotte, and broader interdisciplinary communities within computational creativity, cognitive science, and human-centered AI.
This early period established many of the foundational concepts that would later evolve into co-creative AI, Creative Sense-Making, participatory interaction, and interaction-centered approaches to intelligence. The research focused on distributed cognition, creativity support systems, perceptual interaction, collaborative creativity, and the role of computational systems within dynamic creative environments.
Rather than treating creativity as an isolated mental process occurring solely inside individuals, this work explored creativity as an emergent property of interaction distributed across people, environments, tools, media systems, workflows, perceptual structures, and evolving constraints. Drawing from distributed cognition, human-computer interaction, computational creativity, and cognitive science, the research investigated how computational systems could actively participate in creative processes by shaping interaction, perception, exploration, and coordination rather than merely automating production.
This period also marked an important conceptual shift away from viewing computational systems as passive tools or autonomous generators. Instead, the work increasingly explored how computational systems could function as creativity support partners and eventually as co-creative collaborators engaged in shared creative processes with humans. Many of the conceptual seeds for later frameworks — including artistic computer colleagues, participatory sense-making, quantified co-creation, perceptual logic, and interaction-centered intelligence — emerged during this formative phase.
Importantly, these projects helped establish an interaction-centered perspective on creativity and cognition years before human-AI co-creation became a major topic within contemporary AI discourse. The work emphasized that creativity does not emerge solely from isolated intelligence, but through ongoing interaction between humans, computational systems, materials, environments, and evolving situations. This interaction-centered framing would later become a defining principle throughout the broader Co-Creative AI research program and subsequent work in enactive AI and adaptive interaction systems.
Awarded Best Student Paper at ACM Creativity & Cognition 2011, this work explored how cognition and creativity distribute across people, media systems, workflows, and collaborative environments during filmmaking. The paper helped establish an interaction-centered perspective on creativity.
Introduced early ideas surrounding perceptual logic and collaborative computational art systems, laying conceptual foundations for later work on perceptual attunement and interaction-centered creativity.
Explored how computational systems can scaffold creativity and lower barriers to creative participation for novice creators. This work emphasized creativity support over autonomous generation.
Proposed that creativity support systems should be grounded in cognitive theories of interaction and exploration rather than static productivity models alone. This work helped bridge creativity research and human-computer interaction.
One of the earliest formal articulations of co-creative AI. The paper proposed that computational systems could function as creative collaborators rather than merely tools or autonomous generators. This work helped establish the conceptual foundations of human-AI co-creation.
This interaction-centered framing of co-creativity would later influence broader research directions within computational creativity, co-creative systems, and human-AI collaboration research. Subsequent frameworks across the field increasingly shifted attention from isolated generative capability toward interaction dynamics, collaborative participation, mixed-initiative systems, and the regulation of shared creative processes between humans and AI systems.
This period marked a major theoretical and technical transition in the development of co-creative AI research. Building upon earlier work in distributed creativity and creativity support systems, this era introduced enactive cognition, ecological psychology, participatory interaction, and improvisational collaboration directly into computational creativity research. Rather than treating creativity as the production of isolated outputs generated internally by humans or machines, the work increasingly explored creativity as a dynamic process emerging through continuous interaction between coupled participants situated within evolving environments.
This research challenged the dominant information-processing assumptions that were common throughout both traditional artificial intelligence and computational creativity research at the time. Earlier computational creativity systems often emphasized symbolic generation, planning, search procedures, or autonomous artifact production. In contrast, the enactive perspective proposed that cognition and creativity are fundamentally enacted through ongoing feedback loops between agents and their environments. Creativity was reframed as improvisational, situated, embodied, relational, and dynamically adaptive rather than purely representational or goal-driven.
A central contribution of this era was the introduction of the concept of “artistic computer colleagues.” Rather than designing computational systems as passive creativity support tools or autonomous generators operating independently of human users, this work proposed that AI systems could function as collaborative partners engaged in reciprocal interaction with humans during the creative process itself. These systems were designed to perceive, interpret, respond to, and improvise with human contributions in real time, forming dynamically coupled creative relationships with users.
This period also introduced early formulations of several concepts that would later become foundational throughout the broader Co-Creative AI research ecosystem, including:
interaction-centered creativity
participatory interaction
perceptual logic
improvisational co-creation
dynamically coupled creative systems
real-time creative feedback loops
and interaction-centered models of intelligence.
Importantly, the systems developed during this era were not merely conceptual thought experiments. They became operational research platforms for studying how meaning, coordination, improvisation, and collaborative creativity emerge during human-AI interaction. The work increasingly positioned co-creative systems as experimental environments for investigating cognition itself through interaction. These ideas would later evolve into Creative Sense-Making, quantified co-creation, participatory sense-making research, and broader frameworks for interaction-centered intelligence and enactive AI.
Presented at ICCC 2014, this paper introduced the enactive model of creativity and proposed the concept of “artistic computer colleagues.” The framework synthesized enactive cognition, ecological psychology, improvisation, and participatory interaction into a new paradigm for co-creative AI.
Extended the enactive framework formally into co-creative AI theory. This work proposed that creativity emerges through interaction between dynamically coupled participants rather than isolated symbolic computation alone. The framework argued that creative cognition unfolds through continuous interaction with environments, materials, and collaborators, emphasizing improvisation, perception-action feedback loops, and adaptive coordination as central components of creativity.
One of the earliest true co-creative AI systems. The Drawing Apprentice collaborated with users in real time on a shared drawing canvas through reciprocal improvisational interaction. Rather than generating isolated outputs independently, the system continuously interpreted human contributions and responded dynamically through collaborative drawing behavior. The system became both a co-creative drawing partner and an experimental platform for studying human-AI collaboration, participatory interaction, and emergent co-creative behavior.
Applied enactive cognitive theory to pretend play and imaginative interaction, helping extend interaction-centered theories of cognition and creativity into collaborative play environments. This work further expanded the view that meaning and imaginative engagement emerge dynamically through coordinated interaction rather than solely through internal symbolic reasoning.
This period marked an important transition from theoretical frameworks for co-creative AI toward empirical investigations of interaction itself as a site of cognition, meaning-making, and collaborative creativity. Building upon earlier work in enactive cognition and artistic computer colleagues, the research increasingly focused on how humans and AI systems dynamically coordinate actions, negotiate meaning, improvise together, and sustain collaborative interaction over time. Rather than evaluating creativity solely through final artifacts or isolated outputs, this work examined the unfolding interaction dynamics through which collaborative creativity emerges in situ.
A major conceptual contribution of this era was the introduction of participatory sense-making into co-creative AI research. Drawing from enactive cognitive science, participatory sense-making proposes that meaning does not arise independently inside isolated agents, but emerges relationally through interaction between dynamically coupled participants. This research extended those ideas into human-AI creative systems, exploring whether computational agents could participate meaningfully in collaborative creative processes through reciprocal interaction, adaptive coordination, turn-taking, and improvisational engagement. (dl.acm.org)
This work helped shift co-creative AI away from static notions of human-computer interaction toward a more interaction-centered perspective in which the collaborative process itself became the primary object of study. The research increasingly emphasized:
real-time interaction dynamics
reciprocal feedback loops
emergent coordination
collaborative improvisation
interaction rhythms
dialogical exchange
and the temporal evolution of shared meaning during co-creative activity.
Importantly, these projects also helped establish co-creative systems as empirical research platforms for studying interaction-centered cognition. Systems such as the Drawing Apprentice were no longer viewed merely as artistic applications or creativity support tools. Instead, they became experimental environments for investigating how meaning, creativity, coordination, and participation emerge dynamically during interaction between humans and AI systems. This interaction-centered framing would later contribute directly to subsequent work on Creative Sense-Making, quantified co-creation, creative trajectories, sense-making curves, explainable co-creative AI, and broader theories of interaction-centered intelligence. (dl.acm.org)
This era also introduced increasingly sophisticated computational approaches for modeling interaction itself, including machine learning methods for interpreting embodied motion trajectories, sketch behavior, and dynamic collaborative patterns. The research began bridging co-creative AI with broader investigations into adaptive interaction systems, embodied cognition, and interaction-centered approaches to artificial intelligence more generally.
One of the earliest empirical investigations of participatory sense-making in a human-AI creative system. The paper demonstrated that collaborative meaning-making can emerge dynamically through interaction between humans and AI systems. This work became foundational to interaction-centered co-creative AI research.
Extended the Drawing Apprentice with deep learning-based object recognition to support richer improvisational collaboration. The work explored dialogical interaction between human sketch input and AI interpretation in real time.
Investigated machine learning approaches for modeling embodied motion trajectories and dynamic interaction patterns, contributing to broader research on interaction-centered AI systems.
This period marked a major expansion of co-creative AI research from participatory interaction toward the computational modeling and quantification of collaborative creativity itself. Building upon earlier work in enactive cognition, participatory sense-making, and interaction-centered creativity, this era introduced frameworks for studying how meaning, coordination, creativity, and collaborative dynamics evolve continuously through time during human-AI interaction. Rather than evaluating only final creative artifacts or isolated system outputs, the research increasingly focused on the unfolding interaction process itself as the primary site of creativity and cognition.
A central contribution of this period was the introduction of Creative Sense-Making (CSM), a cognitive framework for modeling co-creative interaction dynamics computationally. Drawing from enactive cognition, participatory sense-making, human-computer interaction, and computational creativity, Creative Sense-Making proposed that collaborative creativity emerges through evolving patterns of interaction between dynamically coupled participants. Creativity was reframed not as a static property of artifacts or isolated individuals, but as an emergent process unfolding through reciprocal interaction, coordination, improvisation, and adaptive sense-making over time. (dl.acm.org)
This work introduced several foundational concepts that would become highly influential within later co-creative AI research, including:
activity traces
creative trajectories
sense-making curves
quantified interaction dynamics
temporal models of collaboration
interaction-centered evaluation
and computational models of collaborative emergence.
A particularly important contribution of this era was the shift toward modeling interaction continuously rather than evaluating collaboration only at discrete endpoints. Traditional computational creativity research often focused on assessing the novelty, value, or quality of completed artifacts. In contrast, this work emphasized the temporal structure of co-creation itself — how collaborative meaning evolves moment-by-moment through sequences of actions, responses, adaptations, interruptions, divergences, recoveries, and mutual influence between participants. (dl.acm.org)
The introduction of creative trajectories and sense-making curves represented one of the earliest attempts to computationally model co-creative interaction as an evolving dynamic system. These frameworks allowed interaction patterns to be visualized and analyzed continuously during collaboration, helping reveal how creativity emerges through interaction rhythms, shifts in participation, conceptual divergence, convergence, and adaptive coordination. This work helped establish a new subfield of co-creative AI research focused on quantifying interaction dynamics in situ rather than evaluating isolated outcomes after collaboration had already ended.
During this period, the research also increasingly integrated deep learning methods into co-creative systems, exploring how machine learning architectures could support conceptual shifts, improvisational interaction, adaptive response generation, and collaborative creativity. Rather than treating AI systems as static generators, the work investigated how computational agents could dynamically influence the trajectory of collaborative interaction itself by provoking novelty, introducing conceptual divergence, and participating in the evolving creative process. These ideas would later contribute directly to explainable co-creative AI, interaction-centered intelligence, adaptive regulation research, and broader theories of enactive AI. (arxiv.org)
Importantly, this era also helped redefine how co-creative systems should be evaluated. Instead of focusing solely on artifact quality, the research proposed that evaluation should account for:
interaction dynamics
user experience
timing of interventions
participatory engagement
collaborative emergence
adaptive coordination
and the evolving structure of interaction through time.
This interaction-centered perspective on evaluation became increasingly influential throughout subsequent work in co-creative AI and human-AI collaboration research more broadly.
Presented at ACM Creativity & Cognition 2017, this paper introduced Creative Sense-Making (CSM), a cognitive framework for modeling and quantifying co-creative interaction dynamics through time. Major contributions included:
activity traces,
creative trajectories,
and:
which modeled collaboration continuously during interaction itself.
The doctoral dissertation expanded the Creative Sense-Making framework into a comprehensive theory of quantified co-creation grounded in enactive cognition, participatory sense-making, and interaction dynamics.
Extended the Creative Sense-Making framework into practical quantified collaboration systems capable of modeling co-creative interaction dynamically through time.
Explored machine learning architectures capable of simultaneously classifying and generating collaborative creative responses within co-creative systems.
Introduced computational methods for generating conceptual shifts during co-creative drawing interaction using deep learning. The work explored how AI systems can intentionally provoke creative divergence and novelty. (arXiv)
Presented a foundational framework for evaluating co-creative systems by focusing on interaction dynamics, user experience, timing of evaluation, and collaborative emergence rather than isolated output quality alone. (arXiv)
Extended conceptual shift modeling into design creativity systems, exploring how AI systems can dynamically provoke more creative outcomes through adaptive collaborative interaction. (arXiv)
Recent work increasingly generalized earlier research in co-creative AI, Creative Sense-Making, participatory interaction, and quantified collaboration into broader theories of interaction-centered artificial intelligence, enactive AI, participatory intelligence, and human-AI co-creation. Building upon more than a decade of interdisciplinary work spanning computational creativity, cognitive science, human-computer interaction, adaptive systems, and enactive cognition, this period increasingly reframed artificial intelligence itself through the lens of interaction, participation, and dynamic sense-making rather than isolated computation or autonomous generation alone.
A central theme of this era is the growing recognition that intelligence, creativity, meaning, and adaptation emerge relationally through ongoing interaction between humans, AI systems, and evolving environments. Earlier work on artistic computer colleagues, participatory sense-making, creative trajectories, and quantified co-creation increasingly expanded into broader theories of interaction-centered intelligence in which cognition is understood as a dynamic process of coordination, regulation, perception-action coupling, and adaptive participation unfolding across time. Rather than conceptualizing AI systems as isolated generators producing outputs independently of users, the research increasingly explored AI systems as dynamically coupled participants engaged in shared cognitive and creative processes with humans. (link.springer.com)
This period also marked a major formalization of enactive AI as a broader theoretical framework for understanding interaction-centered artificial intelligence. Drawing from enactive cognition, ecological psychology, participatory sense-making, adaptive systems theory, and embodied cognition, the work increasingly proposed that intelligence cannot be fully understood through static representations, isolated symbolic computation, or autonomous optimization alone. Instead, intelligence emerges through the ongoing regulation of interaction between agents and environments as systems adapt, coordinate, improvise, and maintain coherence under changing conditions. These ideas increasingly connected earlier co-creative AI research with broader questions involving:
adaptive regulation
participatory intelligence
embodied interaction
interaction-centered cognition
explainability
hybrid intelligence
and the future of human-centered AI systems.
Importantly, this era extended earlier co-creative AI research into the rapidly evolving landscape of generative AI and artificial media. As large-scale generative systems became increasingly capable of producing text, images, music, and multimedia artifacts autonomously, the research increasingly emphasized the importance of preserving visibility into the collaborative interaction processes underlying human-AI creativity. Rather than treating generative AI systems as opaque black boxes producing outputs independently of human participation, this work proposed explainable and quantified co-creative systems capable of modeling how collaborative meaning, authorship, creativity, and interaction evolve dynamically through time during human-AI engagement. (link.springer.com)
This period also introduced a growing emphasis on adaptive regulation and structural drift as central challenges for future AI systems. Earlier work on participatory interaction and quantified collaboration increasingly evolved into broader theories concerning how intelligent systems sustain meaningful interaction over time under changing environmental conditions. The research began exploring interaction-centered intelligence as a general paradigm in which:
intelligence emerges through adaptive participation,
cognition unfolds through interaction,
creativity emerges through collaboration,
and coherent behavior depends upon the continuous regulation of dynamic relationships between agents and environments.
Rather than treating co-creative AI as a niche subfield of computational creativity, this era increasingly positioned human-AI co-creation as an important experimental domain for understanding the future of human-centered artificial intelligence more broadly. Co-creative systems became not only artistic tools or collaborative applications, but empirical research platforms for investigating:
interaction dynamics
participatory cognition
collaborative emergence
adaptive regulation
explainability
and the evolving relationship between humans and intelligent computational systems.
These developments increasingly contributed toward a broader interaction-centered paradigm for artificial intelligence in which participation, adaptation, coordination, and shared sense-making become foundational to understanding intelligence itself.
Published in the Handbook of Human-Centered Artificial Intelligence, this chapter positioned human-AI co-creation as a fundamentally new interaction paradigm centered on participation, adaptation, collaboration, and shared meaning construction. The work synthesized more than a decade of research into interaction-centered AI systems.
Winner of the ICCC 2024 Best Paper Award. This work formalized enactive AI through five pillars:
autonomy,
sense-making,
embodiment,
emergence,
and experience.
The paper established enaction as a foundational framework for understanding co-creative AI systems and interaction-centered intelligence.
Extended the Drawing Apprentice lineage into a modern quantified co-creative AI research platform capable of modeling interaction dynamics, visualizing creative trajectories, and analyzing collaborative interaction using the Creative Sense-Making framework. (arXiv)
Published in Artificial Media: Emerging Trends in Narratives, Education and Creative Practice (Springer), this chapter extends earlier work on Creative Sense-Making, quantified co-creation, and interaction-centered AI into the era of generative artificial intelligence.
The work argues that modern generative AI systems often function as “black boxes” in which the contributions of both human creators and AI systems become obscured within the final artifact. Building upon earlier frameworks involving:
activity traces,
creative trajectories,
participatory interaction,
and quantified collaboration,
the chapter proposes explainable co-creative AI systems capable of revealing and modeling the evolving interaction dynamics underlying human-AI collaboration.
The publication represents an important continuation of earlier co-creative AI research into contemporary concerns involving:
explainability,
authorship,
transparency,
hybrid intelligence,
and human-centered generative AI systems.
Rather than treating AI creativity as isolated autonomous generation, the framework emphasizes:
creativity as an emergent process arising through interaction between humans and AI systems.
Across this body of work, several recurring themes emerge:
Creativity as interaction rather than isolated generation
Intelligence as participatory and relational
Co-creative AI as a collaborative paradigm
Enaction as a foundation for interaction-centered AI
Quantified co-creation through activity traces and creative trajectories
Participatory sense-making between humans and AI systems
Human-AI co-creation as a new interaction paradigm
Together, these works helped establish many of the conceptual foundations now shaping:
co-creative AI,
enactive AI,
human-centered AI,
hybrid intelligence,
participatory AI,
and interaction-centered models of artificial intelligence.