Thomas L. Griffiths

The Riddle Riddle: Testing Flexible Reasoning in Large Language Models and Humans

Bella Fascendini, Kathryn McGregor, Max D. Gupta, et al.

Process Matters more than Output for Distinguishing Humans from Machines

Milena Rmus, Mathew D. Hardy, Thomas L. Griffiths, et al.

Ads in AI Chatbots? An Analysis of How Large Language Models Navigate Conflicts of Interest

Addison J. Wu, Ryan Liu, Shuyue Stella Li, et al.

Serendipity by Design: Evaluating the Impact of Cross-domain Mappings on Human and LLM Creativity

Qiawen Ella Liu, Marina Dubova, Henry Conklin, et al.

This study compares how humans and large language models (LLMs) generate creative ideas, specifically testing whether a technique called 'cross-domain mapping'—forcing creators to draw inspiration from random, unrelated sources—boosts creativity in both. The researchers found that humans benefit significantly from this technique, while LLMs already generate highly original ideas on their own and don't show the same improvement; however, both humans and LLMs do generate more creative ideas when the inspiration source is more different from the target topic.

Parallelograms Strike Back: LLMs Generate Better Analogies than People

Qiawen Ella Liu, Raja Marjieh, Jian-Qiao Zhu, et al.

This study compares how large language models (LLMs) and humans complete word analogies (A:B::C:D problems). The researchers found that LLMs generate better analogies than humans because they more consistently follow the 'parallelogram' geometric structure—where analogies are represented as equal relationships between word pairs in semantic space—while humans often rely on simpler shortcuts like picking familiar words. The results suggest the parallelogram model is actually a good way to understand analogies; humans just struggle to follow it consistently, whereas LLMs naturally do.