CresOWLve: Benchmarking Creative Problem-Solving Over Real-World Knowledge

📰 ArXiv cs.AI

CresOWLve benchmarks creative problem-solving in LLMs using real-world knowledge

advanced Published 7 Apr 2026

Action Steps

Identify the limitations of existing benchmarks for LLMs
Develop a benchmark that evaluates creative problem-solving over real-world knowledge
Use CresOWLve to assess the performance of LLMs in combining logical reasoning, lateral thinking, analogy-making, and commonsense knowledge
Apply the insights from CresOWLve to improve the creative problem-solving capabilities of LLMs

Who Needs to Know This

AI researchers and engineers benefit from this benchmark as it evaluates the creative problem-solving capabilities of LLMs, while product managers can use it to assess the potential of LLMs in real-world applications

Key Insight

💡 CresOWLve provides a comprehensive evaluation of LLMs' creative problem-solving abilities, going beyond traditional benchmarks

Key Takeaways

CresOWLve benchmarks creative problem-solving in LLMs using real-world knowledge

Full Article

Title: CresOWLve: Benchmarking Creative Problem-Solving Over Real-World Knowledge

Abstract:
arXiv:2604.03374v1 Announce Type: cross Abstract: Creative problem-solving requires combining multiple cognitive abilities, including logical reasoning, lateral thinking, analogy-making, and commonsense knowledge, to discover insights that connect seemingly unrelated pieces of information. However, most existing benchmarks for large language models (LLMs) evaluate only specific components of this process. Moreover, many creativity-oriented benchmarks rely on artificially constructed brainteasers

Read full paper → ← Back to Reads