Trend AnalysisArts & Design
Computational Creativity and Algorithmic Composition: When Machines Become Co-Authors
Computational creativity has moved beyond generating plausible outputs to raising fundamental questions about what creativity is. New frameworks for Generative Collective Intelligence and algorithmically literate art criticism suggest the field is maturing from technical demos to cultural practice.
By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.
Why It Matters
Computational creativityโthe study and development of software that exhibits behavior that would be considered creative in humansโhas existed as a research field for decades. But the emergence of large generative models has transformed it from a niche academic pursuit into a cultural force. When AI can generate poetry, music, visual art, and narrative that audiences find meaningful, the question shifts from "can machines be creative?" to "what does creativity mean when machines participate in it?"
This shift has consequences for artistic practice, education, and criticism. If creativity is no longer an exclusively human capacity, then the frameworks through which we evaluate, teach, and reward creative work must be fundamentally reconsidered. Recent research suggests this reconceptualization is already underwayโand it is more nuanced than either the techno-utopian or techno-dystopian narratives suggest.
The Science / The Practice
Machine-Led Aesthetics
Paksi (2025) presents a narrative review examining how human-AI collaboration reshapes creativity, emphasizing both opportunities and ethical tensions. The paper introduces the concept of "machine-led aesthetics"โaesthetic qualities that emerge from algorithmic processes rather than human intention. This is a provocative framing: rather than asking whether AI can produce art that meets human aesthetic standards, Paksi asks whether AI creates new aesthetic categories that humans learn to appreciate. The parallel to photography's early historyโinitially dismissed as non-art, now a fully recognized mediumโis instructive.
Generative Collective Intelligence
Kehler and Pentland (2025), with 4 citations, propose a framework called Generative Collective Intelligence (GCI) that positions AI in dual roles: as interactive agents and as infrastructure that accumulates, organizes, and leverages collective human knowledge. The framework, emerging from MIT and Santa Fe Institute research, argues that the most powerful creative outcomes arise not from individual human-AI pairs but from systems where AI mediates collaboration among many humans. This represents a shift from "AI as creative tool" to "AI as creative ecosystem"โa fundamentally different vision of computational creativity.
Algorithmically Literate Art Criticism
Maheswari et al. (2025) argue that AI-generated art requires a new form of art criticismโone that is "algorithmically literate." Traditional art criticism evaluates works based on human intention, cultural context, emotional expression, and technical skill. AI-generated art challenges all four criteria. The paper proposes a framework for evaluating AI art that incorporates understanding of the generative process, training data provenance, and the human-AI interaction that shaped the output. This is a significant contribution to arts education and curatorial practice.
Emergent Creativity from Simple Rules
Schaap and Hedblom (2024) provide a fascinating case study with AutomaTone, an interactive music generator based on Conway's Game of Life. The system produces music through cellular automataโsimple rules that generate complex emergent behavior. The paper discusses whether such emergent outputs can be called "creative," connecting computational creativity to complexity theory and emergence. The finding is that creativity may not require intention: complex, aesthetically interesting outputs can emerge from systems with no goals, preferences, or awarenessโa philosophically challenging result.
Computational Creativity Framework Comparison
<
| Framework | Core Idea | Human Role | AI Role |
|---|
| AI as tool | AI extends human capability | Director, curator | Instrument |
| Machine-led aesthetics (Paksi) | AI creates new aesthetic categories | Appreciator, interpreter | Originator |
| Generative Collective Intelligence (Kehler et al.) | AI mediates collective human creativity | Contributor in collective | Infrastructure |
| Algorithmic criticism (Maheswari et al.) | New evaluation criteria needed | Critic with algorithmic literacy | Subject of evaluation |
| Emergent creativity (Schaap & Hedblom) | Creativity from simple rules | Observer | Emergent system |
What To Watch
The field is moving toward a post-anthropocentric view of creativityโnot denying human creativity's value, but recognizing that creativity may be a property of complex systems rather than an exclusively human trait. Watch for the development of standardized evaluation frameworks for computational creativity (the field currently lacks consensus metrics), and for the integration of GCI-style collective frameworks into creative industries where teams of humans and AI agents collaborate on large-scale projects like film, game development, and architectural design.
Explore related work through ORAA ResearchBrain.
๋ฉด์ฑ
์กฐํญ: ์ด ๊ฒ์๋ฌผ์ ์ ๋ณด ์ ๊ณต ๋ชฉ์ ์ ์ฐ๊ตฌ ๋ํฅ ๊ฐ์์ด๋ค. ํ์ ์์
์์ ์ธ์ฉํ๊ธฐ ์ ์ ํน์ ์ฐ๊ตฌ ๊ฒฐ๊ณผ, ํต๊ณ ๋ฐ ์ฃผ์ฅ์ ์๋ณธ ๋
ผ๋ฌธ์ ํตํด ๊ฒ์ฆํด์ผ ํ๋ค.
์ปดํจํ
์ด์
๋ ํฌ๋ฆฌ์์ดํฐ๋นํฐ์ ์๊ณ ๋ฆฌ์ฆ ์๊ณก: ๊ธฐ๊ณ๊ฐ ๊ณต๋ ์ ์๊ฐ ๋ ๋
์ ์ค์ํ๊ฐ
์ปดํจํ
์ด์
๋ ํฌ๋ฆฌ์์ดํฐ๋นํฐ(computational creativity)โ์ธ๊ฐ์๊ฒ์ ์ฐฝ์์ ์ด๋ผ๊ณ ์ฌ๊ฒจ์ง ํ๋์ ๋ณด์ด๋ ์ํํธ์จ์ด์ ์ฐ๊ตฌ ๋ฐ ๊ฐ๋ฐโ๋ ์์ญ ๋
๊ฐ ์ฐ๊ตฌ ๋ถ์ผ๋ก ์กด์ฌํด ์๋ค. ๊ทธ๋ฌ๋ ๋๊ท๋ชจ ์์ฑ ๋ชจ๋ธ์ ๋ฑ์ฅ์ ์ด๋ฅผ ํ์ ํ๋ฌธ์ ์ถ๊ตฌ์์ ๋ฌธํ์ ํ์ผ๋ก ๋ณ๋ชจ์์ผฐ๋ค. AI๊ฐ ๊ด๊ฐ์ด ์๋ฏธ ์๋ค๊ณ ๋๋ผ๋ ์, ์์
, ์๊ฐ ์์ , ์์ฌ๋ฅผ ์์ฑํ ์ ์๊ฒ ๋๋ฉด์, ์ง๋ฌธ์ "๊ธฐ๊ณ๊ฐ ์ฐฝ์์ ์ผ ์ ์๋๊ฐ?"์์ "๊ธฐ๊ณ๊ฐ ์ฐฝ์์ฑ์ ์ฐธ์ฌํ ๋ ์ฐฝ์์ฑ์ ๋ฌด์์ ์๋ฏธํ๋๊ฐ?"๋ก ์ ํ๋์๋ค.
์ด๋ฌํ ์ ํ์ ์์ ์ ์ค์ฒ, ๊ต์ก, ๋นํ์ ์ํฅ์ ๋ฏธ์น๋ค. ์ฐฝ์์ฑ์ด ๋ ์ด์ ์ธ๊ฐ๋ง์ ๋ฅ๋ ฅ์ด ์๋๋ผ๋ฉด, ์ฐ๋ฆฌ๊ฐ ์ฐฝ์์ ์์
์ ํ๊ฐํ๊ณ , ๊ฐ๋ฅด์น๊ณ , ๋ณด์ํ๋ ํ์ ๊ทผ๋ณธ์ ์ผ๋ก ์ฌ๊ณ ํด์ผ ํ๋ค. ์ต๊ทผ ์ฐ๊ตฌ๋ ์ด๋ฌํ ์ฌ๊ฐ๋
ํ๊ฐ ์ด๋ฏธ ์งํ ์ค์์ ์์ฌํ๋ฉฐโ๊ธฐ์ ์ ํ ํผ์์ ๋๋ ๊ธฐ์ ๋์คํ ํผ์์ ์์ฌ๊ฐ ์ ์ํ๋ ๊ฒ๋ณด๋ค ํจ์ฌ ๋ ๋ฏธ๋ฌํ ๋ฐฉ์์ผ๋ก ์ด๋ฃจ์ด์ง๊ณ ์๋ค.
๊ณผํ / ์ค์ฒ
๊ธฐ๊ณ ์ฃผ๋ ๋ฏธํ
Paksi(2025)๋ ์ธ๊ฐ-AI ํ์
์ด ์ฐฝ์์ฑ์ ์ด๋ป๊ฒ ์ฌํธํ๋์ง๋ฅผ ๊ฒํ ํ๋ ์์ ์ ๋ฆฌ๋ทฐ๋ฅผ ์ ์ํ๋ฉฐ, ๊ธฐํ์ ์ค๋ฆฌ์ ๊ธด์ฅ ๋ชจ๋๋ฅผ ๊ฐ์กฐํ๋ค. ์ด ๋
ผ๋ฌธ์ "๊ธฐ๊ณ ์ฃผ๋ ๋ฏธํ(machine-led aesthetics)"์ด๋ผ๋ ๊ฐ๋
์ ๋์
ํ๋คโ์ธ๊ฐ์ ์๋๊ฐ ์๋ ์๊ณ ๋ฆฌ์ฆ ํ๋ก์ธ์ค์์ ๋ฐ์ํ๋ ๋ฏธ์ ํน์ง์ด๋ค. ์ด๋ ๋๋ฐ์ ์ธ ํ๋ ์ด๋ฐ์ด๋ค. AI๊ฐ ์ธ๊ฐ์ ๋ฏธ์ ๊ธฐ์ค์ ์ถฉ์กฑํ๋ ์์ ์ ์์ฐํ ์ ์๋์ง๋ฅผ ๋ฌป๋ ๋์ , Paksi๋ AI๊ฐ ์ธ๊ฐ์ด ํ์ตํ์ฌ ๊ฐ์ํ๊ฒ ๋๋ ์๋ก์ด ๋ฏธ์ ๋ฒ์ฃผ๋ฅผ ์ฐฝ์กฐํ๋์ง๋ฅผ ๋ฌป๋๋ค. ์ด๊ธฐ์ ๋น์์ ๋ก ๋ฌด์๋๋ค๊ฐ ํ์ฌ๋ ์์ ํ ์ธ์ ๋ฐ๋ ๋งค์ฒด๊ฐ ๋ ์ฌ์ง์ ์ด๊ธฐ ์ญ์ฌ์์ ์ ์ฌ์ฑ์ ์์ฌํ๋ ๋ฐ๊ฐ ํฌ๋ค.
์์ฑ์ ์ง๋จ ์ง์ฑ
Kehler์ Pentland(2025)๋ 4ํ ์ธ์ฉ์ผ๋ก, AI๋ฅผ ์ํธ์์ฉํ๋ ์์ด์ ํธ์ด์ ์ง๋จ์ ์ธ๊ฐ ์ง์์ ์ถ์ , ์กฐ์ง, ํ์ฉํ๋ ์ธํ๋ผ๋ก์์ ์ด์ค ์ญํ ์ ์์น์ํค๋ ์์ฑ์ ์ง๋จ ์ง์ฑ(Generative Collective Intelligence, GCI) ํ๋ ์์ํฌ๋ฅผ ์ ์ํ๋ค. MIT์ Santa Fe Institute ์ฐ๊ตฌ์์ ๋์ถ๋ ์ด ํ๋ ์์ํฌ๋, ๊ฐ์ฅ ๊ฐ๋ ฅํ ์ฐฝ์์ ๊ฒฐ๊ณผ๋ฌผ์ด ๊ฐ๋ณ ์ธ๊ฐ-AI ์์ด ์๋๋ผ AI๊ฐ ๋ค์์ ์ธ๊ฐ ๊ฐ ํ์
์ ๋งค๊ฐํ๋ ์์คํ
์์ ๋ฐ์ํ๋ค๊ณ ์ฃผ์ฅํ๋ค. ์ด๋ "์ฐฝ์์ ๋๊ตฌ๋ก์์ AI"์์ "์ฐฝ์์ ์ํ๊ณ๋ก์์ AI"๋ก์ ์ ํ์ ๋ํ๋ด๋ฉฐโ์ปดํจํ
์ด์
๋ ํฌ๋ฆฌ์์ดํฐ๋นํฐ์ ๋ํ ๊ทผ๋ณธ์ ์ผ๋ก ๋ค๋ฅธ ๋น์ ์ด๋ค.
์๊ณ ๋ฆฌ์ฆ์ ์ผ๋ก ์์ ์๋ ์์ ๋นํ
Maheswari ์ธ(2025)๋ AI ์์ฑ ์์ ์ด "์๊ณ ๋ฆฌ์ฆ์ ์ผ๋ก ์์ ์๋(algorithmically literate)" ์๋ก์ด ํํ์ ์์ ๋นํ์ ์๊ตฌํ๋ค๊ณ ์ฃผ์ฅํ๋ค. ์ ํต์ ์ธ ์์ ๋นํ์ ์ธ๊ฐ์ ์๋, ๋ฌธํ์ ๋งฅ๋ฝ, ๊ฐ์ ์ ํํ, ๊ธฐ์ ์ ์ญ๋์ ๊ธฐ๋ฐํ์ฌ ์ํ์ ํ๊ฐํ๋ค. AI ์์ฑ ์์ ์ ์ด ๋ค ๊ฐ์ง ๊ธฐ์ค ๋ชจ๋์ ๋์ ํ๋ค. ์ด ๋
ผ๋ฌธ์ ์์ฑ ํ๋ก์ธ์ค์ ๋ํ ์ดํด, ํ๋ จ ๋ฐ์ดํฐ ์ถ์ฒ, ๊ฒฐ๊ณผ๋ฌผ์ ํ์ฑํ ์ธ๊ฐ-AI ์ํธ์์ฉ์ ํฌํจํ๋ AI ์์ ํ๊ฐ ํ๋ ์์ํฌ๋ฅผ ์ ์ํ๋ค. ์ด๋ ์์ ๊ต์ก ๋ฐ ํ๋ ์ดํ ๋ฆฌ์ผ ์ค์ฒ์ ๋ํ ์ค์ํ ๊ธฐ์ฌ์ด๋ค.
๋จ์ํ ๊ท์น์์ ๋ฐ์ํ๋ ์ฐฝ์์ฑ
Schaap๊ณผ Hedblom(2024)์ Conway์ ์๋ช
๊ฒ์(Game of Life)์ ๊ธฐ๋ฐํ ์ธํฐ๋ํฐ๋ธ ์์
์์ฑ๊ธฐ์ธ AutomaTone์ ๋ํ ํฅ๋ฏธ๋ก์ด ์ฌ๋ก ์ฐ๊ตฌ๋ฅผ ์ ์ํ๋ค. ์ด ์์คํ
์ ์ธํฌ ์๋์(cellular automata)โ๋ณต์กํ ์ฐฝ๋ฐ์ ํ๋์ ์์ฑํ๋ ๋จ์ํ ๊ท์นโ๋ฅผ ํตํด ์์
์ ์์ฐํ๋ค. ์ด ๋
ผ๋ฌธ์ ๊ทธ๋ฌํ ์ฐฝ๋ฐ์ ๊ฒฐ๊ณผ๋ฌผ์ "์ฐฝ์์ "์ด๋ผ๊ณ ๋ถ๋ฅผ ์ ์๋์ง๋ฅผ ๋
ผ์ํ๋ฉฐ, ์ปดํจํ
์ด์
๋ ํฌ๋ฆฌ์์ดํฐ๋นํฐ๋ฅผ ๋ณต์ก๊ณ ์ด๋ก ๋ฐ ์ฐฝ๋ฐ๊ณผ ์ฐ๊ฒฐํ๋ค. ์ฐฝ์์ฑ์๋ ์๋๊ฐ ํ์ํ์ง ์์ ์ ์๋ค๋ ๊ฒ์ด ๋ฐ๊ฒฌ ๊ฒฐ๊ณผ์ด๋ค. ์ฆ, ๋ณต์กํ๊ณ ๋ฏธ์ ์ผ๋ก ํฅ๋ฏธ๋ก์ด ๊ฒฐ๊ณผ๋ฌผ์ ๋ชฉํ, ์ ํธ, ๋๋ ์ธ์์ด ์๋ ์์คํ
์์๋ ์ฐฝ๋ฐํ ์ ์์ผ๋ฉฐโ์ด๋ ์ฒ ํ์ ์ผ๋ก ๋์ ์ ์ธ ๊ฒฐ๊ณผ์ด๋ค.
์ปดํจํ
์ด์
๋ ํฌ๋ฆฌ์์ดํฐ๋นํฐ ํ๋ ์์ํฌ ๋น๊ต
<
| ํ๋ ์์ํฌ | ํต์ฌ ๊ฐ๋
| ์ธ๊ฐ์ ์ญํ | AI์ ์ญํ |
|---|
| ----------- | ----------- | ------------ | --------- |
| ๋๊ตฌ๋ก์์ AI | AI๊ฐ ์ธ๊ฐ์ ๋ฅ๋ ฅ์ ํ์ฅํจ | ๊ฐ๋
์, ํ๋ ์ดํฐ | ๋๊ตฌ |
| ๊ธฐ๊ณ ์ฃผ๋ ๋ฏธํ (Paksi) | AI๊ฐ ์๋ก์ด ๋ฏธํ์ ๋ฒ์ฃผ๋ฅผ ์ฐฝ์ถํจ | ๊ฐ์์, ํด์์ | ์ฐฝ์์ |
| ์์ฑ์ ์ง๋จ ์ง์ฑ (Kehler et al.) | AI๊ฐ ์ง๋จ์ ์ธ๊ฐ ์ฐฝ์์ฑ์ ๋งค๊ฐํจ | ์ง๋จ ๋ด ๊ธฐ์ฌ์ | ์ธํ๋ผ |
| ์๊ณ ๋ฆฌ์ฆ ๋นํ (Maheswari et al.) | ์๋ก์ด ํ๊ฐ ๊ธฐ์ค์ด ํ์ํจ | ์๊ณ ๋ฆฌ์ฆ ๋ฆฌํฐ๋ฌ์๋ฅผ ๊ฐ์ถ ๋นํ๊ฐ | ํ๊ฐ ๋์ |
| ์ฐฝ๋ฐ์ ์ฐฝ์์ฑ (Schaap & Hedblom) | ๋จ์ํ ๊ท์น์ผ๋ก๋ถํฐ ์ฐฝ์์ฑ์ด ์ถํํจ | ๊ด์ฐฐ์ | ์ฐฝ๋ฐ์ ์์คํ
|
์ฃผ๋ชฉํ ๋ํฅ
์ด ๋ถ์ผ๋ ์ฐฝ์์ฑ์ ๋ํ ํ์ธ๊ฐ์ค์ฌ์ ๊ด์ ์ผ๋ก ๋์๊ฐ๊ณ ์๋คโ์ด๋ ์ธ๊ฐ ์ฐฝ์์ฑ์ ๊ฐ์น๋ฅผ ๋ถ์ ํ๋ ๊ฒ์ด ์๋๋ผ, ์ฐฝ์์ฑ์ด ์ธ๊ฐ๋ง์ ๊ณ ์ ํ ํน์ฑ์ด ์๋ ๋ณต์ก๊ณ์ ์์ฑ์ผ ์ ์์์ ์ธ์ํ๋ ๊ฒ์ด๋ค. ์ปดํจํ
์ด์
๋ ์ฐฝ์์ฑ์ ๋ํ ํ์คํ๋ ํ๊ฐ ํ๋ ์์ํฌ์ ๊ฐ๋ฐ(ํ์ฌ ์ด ๋ถ์ผ์๋ ํฉ์๋ ์งํ๊ฐ ์๋ค)๊ณผ, ์ํยท๊ฒ์ ๊ฐ๋ฐยท๊ฑด์ถ ์ค๊ณ ๋ฑ ๋๊ท๋ชจ ํ๋ก์ ํธ์์ ์ธ๊ฐ๊ณผ AI ์์ด์ ํธ ํ์ด ํ์
ํ๋ ์ฐฝ์ ์ฐ์
์ GCI ๋ฐฉ์์ ์ง๋จ ํ๋ ์์ํฌ๊ฐ ํตํฉ๋๋ ํ๋ฆ์ ์ฃผ๋ชฉํ ํ์๊ฐ ์๋ค.
๊ด๋ จ ์ฐ๊ตฌ๋ ORAA ResearchBrain์์ ํ์ํ ์ ์๋ค.
References (5)
[1] Paksi, D. N. F. (2025). Beyond Human Authorship: Exploring Computational Creativity and Machine-Led Aesthetics. Harmonia, 3(3).
[2] Kehler, T. P., Page, S., & Pentland, A. (2025). Amplifying Human Creativity and Problem Solving with AI Through Generative Collective Intelligence. arXiv.
[3] Maheswari, M., Raj, N., & Sakthivel, E. (2025). Reinventing Art Criticism in the AI Era. ShodhKosh.
[4] Schaap, G., & Hedblom, M. M. (2024). Discussing the Creativity of AutomaTone: an Interactive Music Generator based on Conway's Game of Life. ICCC Workshop.
M. Maheswari, Raj, N., E Sakthivel, Dalei, V., Sidhu, A., & Pradhan, I. P. (2025). REINVENTING ART CRITICISM IN THE AI ERA. ShodhKosh: Journal of Visual and Performing Arts, 6(2s).