Collective creativity in hybrid societies
Mason Youngblood, Katie Mudd, Manuel Anglada-Tort, Cameron Jones, Elena Miu, Diana Omigie, Margaret Schedel
cs.AI · cs.CY · cs.MA
Quick verdict
Position paper arguing creativity with AI is a property of human-AI collectives; persuasive synthesis, but key evidence rests on unpublished preprints.
Summary
This is a position/review paper arguing that creativity in the age of generative AI should be understood as an emergent property of 'hybrid collectives' — populations of interacting humans and algorithms — rather than of individuals. The authors open by separating two constructs the literature conflates: novelty (a property of a single artifact) and diversity (a property of a set). They then organize mechanisms of idea generation (combination, search, transformation; Table 1) around the cultural-evolution cycle of variation, selection, and retention, reviewing evidence that AI-assisted ideation reliably raises individual-output novelty while narrowing population diversity (e.g., 94% of ChatGPT-assisted brainstorming ideas shared overlapping concepts). Synthesizing transmission-chain experiments, networked search studies, and platform data (130,882 Artbreeder remix lineages; four centuries of Go records), they argue mixed human-AI groups can outperform and out-diversify homogeneous ones, making composition — who is present, in what proportion, how connected — the decisive variable. Sections on evaluation, bias amplification (53% to 65% in AI-mediated networks), ownership, and environmental costs broaden the scope. Main limitations: no original data; the central supporting studies (hybrid-chain superiority, strategy propagation) are unpublished 2026 preprints, several from the authors' own labs; and the population-level diversity metrics the argument requires do not yet exist. It closes with a three-part agenda: population-scale designs, composition as independent variable, and standardized measurement.
Review complete
Assessment at a glance
4
risk flags identified
● high 0
● medium 2
● low 2
5 strengths · 5 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Creativity in the age of generative AI is best understood as an emergent property of hybrid collectives of interacting humans and algorithms, not of individuals.
- The apparent disagreement over whether generative AI enriches or impoverishes culture stems from conflating novelty (an artifact-level property) with diversity (a population-level property).
- AI-assisted ideation reliably raises the novelty of individual output while narrowing aggregate diversity, primarily because many people sample from the same generative distribution.
- Mixed human-AI groups can outperform and out-diversify homogeneous groups because humans and models search in complementary ways, making group composition the decisive variable.
- Convergence is partly a design choice rather than an inevitable consequence of AI, and machine-discovered strategies can enter and persist in human culture when seeded into it.
Strengths
The novelty/diversity distinction dissolves an apparent contradiction
Sections 1-2 cleanly separate artifact-level novelty from population-level diversity, reconciling 'AI boosts creativity' findings (refs 32-34) with 'AI homogenizes' findings (refs 33, 35) that otherwise look inconsistent — the conceptual core of the paper and its most transferable contribution.
Coherent organizing framework with an honest mechanism table
Table 1 cross-maps three generation mechanisms (combination, search, transformation) onto what humans, machines, and hybrids each contribute, candidly flagging transformation as where machines contribute least (refs 31, 36, 39) rather than inflating AI capability.
Respects the evidence hierarchy
Scope includes quantified, neglected consequences
Actionable, self-aware research agenda
Weaknesses
Central claims rest on unpublished preprints, several self-authored
No original data or re-analysis
The complementarity mechanism is supported thinly
Asymmetric domain coverage
Normative dimensions are catalogued, not integrated
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Scope & Claims
This is a position paper — a conceptual synthesis, not an empirical study — making two reframings. First, the literature's apparent contradiction (AI tools raise individual creativity yet homogenize culture) dissolves once **novelty** (a property of a single artifact, measured against a reference set) is separated from **diversity** (a property of a set: how much its members differ). Second, creativity in the age of generative AI is an emergent property of *hybrid collectives* — populations of interacting humans and algorithms — so the operative question becomes composition (which agents, in what proportion, how connected), not whether AI helps or harms. The paper organizes mechanisms (combination, search, transformation; Table 1) around the cultural-evolution cycle of variation, selection, retention (Section 3), illustrates across science, arts, and games (Section 4), and derives consequences plus a research agenda (Sections 5-6).
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 5 strengths & 5 weaknesses, analyzed in detail
- 4 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
What is a hybrid collective in the context of AI and creativity?
In this paper, a hybrid collective is a population of interacting humans and AI systems viewed as a single creative system. The authors argue that creativity in the age of generative AI is a property of such collectives — which agents are present, in what proportion, and how they are connected — rather than of any individual person or model.
What is the difference between novelty and diversity in creativity research?
Novelty is a property of a single artifact: how far it departs from prior work, measured against a reference set. Diversity is a property of a set of artifacts: how much the members differ from one another. The paper argues generative AI tools tend to raise novelty (each output looks more original) while lowering diversity (many people's outputs resemble each other), and that conflating the two explains much of the disagreement about AI and creativity.
Does AI make creative output more or less diverse?
According to the studies reviewed, both, at different levels: individuals using AI produce more novel work, but when many people use the same model the pooled output becomes less diverse — in one study, 94% of ChatGPT-assisted brainstorming ideas shared overlapping concepts. Mixed human-AI groups, however, can preserve more diversity than humans-only or AI-only groups, so the outcome depends on group composition.
Is this paper an empirical study?
No. It is a position/review paper that synthesizes existing experiments, platform datasets, and case studies — transmission chains, Artbreeder remix data, Go game records, book-market analyses — into a conceptual framework. It reports no new data and no statistical analysis of its own.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Superscript citation markers were flattened into running text by PDF extraction (e.g., 'creativity1', 'idea generation1'); citation numbers were reconstructed from the reference list and context. The extracted text spans the complete document from abstract through reference 98 with no truncation.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.