top of page

What makes interaction meaningful? How do we generate new knowledge via interaction between agents? How can language-based interaction change behavior or elicit transformative insights?

RESEARCH & DISCOVERIES

human-AI interaction triangle

At our lab, we believe that universal principles govern meaningful interactions. Thus, we developed a triangular framework addressing human-to-human, human-to-model, and model-to-model meaningful interactions. 

In our research, we use large language models (LLMs) and other computational modeling methods to explore human and synthetic cognitions and build computational models of human and artificial minds. 
brain connections

LAB NEWS

Thrilled to share that our paper "Systematic Biases in LLM Simulations of Debates" has been accepted to #EMNLP2024! 🎉

We demonstrate that LLM agents adhere to the model's inherent social biases, even if these biases conflict with their assigned identities. We enforce these findings using our self-fine-tuning method, which can manipulate LLMs’ intrinsic biases effectively.

AVAILABLE COURSES

Ariel Goldstein teaches courses both in the Cognitive Science department and in the Business School in the Hebrew University. These are the courses offered this year:

​

1) Between Artificial and Human Intelligence

​

2) Data Science Practicum

ONLINE LECTURES

bottom of page