Creative Priming: Decoupling Process from Product to Escape the Creative Paradox
cs.AI, cs.CL
Submitted: 2026-11-09
Updated: 2026-11-09
Code: https://github.com/deedy/imo-2026
License: http://creativecommons.org/licenses/by/4.0/
The gist: Many of the most consequential advances promised by AI, from curing diseases and inventing new technologies to formulating new scientific theories, are fundamentally creative acts.
Terminology
Abstract
Many of the most consequential advances promised by AI, from curing diseases and inventing new technologies to formulating new scientific theories, are fundamentally creative acts. Yet creativity is uniquely difficult to learn because it faces a creative paradox: learning to repeat past examples of creative success makes those specific examples more likely to recur, inhibiting the very creativity desired. This work argues that creativity is better modeled as an explicit process rather than as a property of individual artifacts, and proposes to learn the underlying generative principles behind the creative process. In particular, to escape the paradox, we introduce creative priming (CP), wherein a trained priming model generates a prompt-specific primer that assists a response model in being creative, explicitly decoupling the creative process from response generation. The priming model learns through GRPO, rewarding primers that lead to responses that are novel, high-quality, and diverse, thereby reinforcing the process that yields creative responses without reinforcing any specific response.
Sources
- Heuresis: Search Strategies for Autonomous AI Research Agents Across Quality, Diversity and Novelty
- HealthBench: Evaluating Large Language Models Towards Improved Human Health
- Vector Policy Optimization: Training for Diversity Improves Test-Time Search
- MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering
- Measuring the Gap Between Human and LLM Research Ideas
- SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?
- The Vendi Score: A Diversity Evaluation Metric for Machine Learning
- Training AI Co-Scientists Using Rubric Rewards
- GIANTS: Generative Insight Anticipation from Scientific Literature
- Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond)
- The Road Less Traveled: Enhancing Exploration in LLMs via Sequential Sampling
- Jointly Reinforcing Diversity and Quality in Language Model Generations
- HypoBench: Towards Systematic and Principled Benchmarking for Hypothesis Generation
- AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agents
- Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces
- MLGym: A New Framework and Benchmark for Advancing AI Research Agents
- GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks
- GPQA: A Graduate-Level Google-Proof Q&A Benchmark
- Artificial Intelligence Index Report 2026
- Recovering Diversity Without Losing Alignment: A DPO Recipe for Post-Trained LLMs
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents