Samuel Schapiro

I am an interdisciplinary scholar studying the science of intelligence, affiliated with the Knowledge Lab at the University of Chicago and the AI Innovation Institute (AI³) at Stony Brook University, and closely collaborating with the Department of Psychology at Pennsylvania State University. My work bridges cognitive science, computer science, and the philosophy of science to advance our understanding of creativity, our ability to elicit it in intelligent machines, and its applications towards scientific, industrial, and artistic domains.

Previously, I was a visiting researcher at the Simons Institute for the Theory of Computing, UC Berkeley. I did my undergrad at the University of Illinois, Urbana-Champaign, where I studied theoretical computer science.

Email / Google Scholar / GitHub / LinkedIn / Bookshelf

News

[Jul 2026] 2/2 papers accepted at COLM 2026; TwistBench accepted as an Oral at the COLM 2026 Workshop on Scientific Understanding of Foundation Models.
[May 2026] 2/2 papers accepted at the ICML 2026 Workshop on Generative AI & Creativity.
[Apr 2026] Received the Lambda AI Research Grant.
[Jun 2025] Best Short Paper Award at ICCC 2025 for Transformational Creativity in Science: A Graphical Theory.
[May 2025] 2/2 papers accepted at ICCC 2025.

A post by @samschapiro on X

Research Areas

My work is organized around three interrelated topics. For a complete list of publications, see my Google Scholar.

I. Measuring the Creativity of AI

Semantic distance tests plotted against creativity benchmarks

UMAP of literary plot-twist clusters from TwistBench

Benchmarks and tests of creativity, especially more complex forms such as transformational and scientific creativity.

TwistBench: Benchmarking Transformational Creativity in LLMs via Literary Plot Twists S. Schapiro, A. Gladstone, H. Ji, R. E. Beaty Oral, COLM 2026 Workshop on Scientific Understanding of Foundation Models.
Assessing the Creativity of Large Language Models: Testing, Limits, and New Frontiers S. Schapiro, A. Gladstone, J. Black, H. Ji Preprint
Do Semantic Distance Tests Actually Predict Creativity in Large Language Models? S. Schapiro, A. Gladstone, J. Black, H. Ji Accepted at the ICML 2026 Workshop on Generative AI & Creativity. Under review, NeurIPS.

II. Improving the Creativity of AI

CreativityNeuro weight steering and its effect on generation diversity

Spark system diagram: Xplor, idea generator, and filter stages

Methods and systems that make models more creative.

CreativityNeuro: Steering Language Model Weights to Improve Divergent Thinking and Reduce Mode Collapse S. Schapiro, C. Francisco Park, F. Sosa, L. R. Varshney Third Conference on Language Modeling (COLM). Accepted at the ICML 2026 Workshop on Generative AI & Creativity.
Spark: A System for Scientifically Creative Idea Generation A. Sanyal, S. Schapiro, S. Shashidhar, R. Moon, L. R. Varshney, D. Hakkani-Tur Proceedings of the 16th International Conference on Computational Creativity (ICCC)

III. Theories and Foundations of Creativity

Graphical theory of transformational creativity in science

Combinatorial creativity: conceptual graphs behind ordinary and extraordinary insight

Formal accounts of what creativity is and how it manifests in minds and machines.

Artificial Creative Intelligence S. Schapiro, R. E. Beaty Invited Chapter, Cambridge Handbook of Creativity (3rd ed.), Cambridge University Press. Eds. J. C. Kaufman & R. J. Sternberg. In preparation.
Combinatorial Creativity: A New Frontier in Generalization Abilities S. Schapiro, S. Shashidhar, A. Gladstone, J. Black, R. Moon, D. Hakkani-Tur, L. R. Varshney Preprint
Transformational Creativity in Science: A Graphical Theory S. Schapiro, J. Black, L. R. Varshney Proceedings of the 16th ICCC — Best Short Paper Award

Service

Conference Reviewer: ICML 2026, COLM 2026, NeurIPS 2026, AAAI 2026
Journal Reviewer: Proceedings of the National Academy of Sciences (PNAS)
Workshop Reviewer: NeurIPS 2024 Creativity & Generative AI, ICML 2026 Creativity & Generative AI