Can Machines Learn Like Humans?
Artificial intelligence has pervaded our lives, from computer vision to ChatGPT, but how effectively can machines emulate human thought? Some machines have gotten impressively good at predicting outcomes, yet the ability to truly reason and ask “what if” remains elusive. In this pair of talks and moderated discussion, three experts spanning the fields of cognitive science and machine learning come together to discuss the next frontier at the intersection of natural and artificial intelligence.
Zenna Tavares, PhD, Alan Kanzer Innovation Scholar and Associate Research Scientist at Columbia University’s Zuckerman Institute and Data Science Institute and Co-Founder and Director of Basis Research Institute, will open our event by sharing his work on a fundamental yet complex question: how do we make sense of the world? By studying how our brains reason, he explores what would it take for a machine to “think” like, or even better than, a human.
Kimberly Stachenfeld, PhD, Senior Research Scientist at Google DeepMind and Affiliate Faculty at the Center for Theoretical Neuroscience at Columbia University, will then speak about her research deciphering the math behind how the mind works. The brain acts as a simulator for our mental experience of the world, but can a computer model faithfully capture its essence? In what ways can a deeper understanding of this process unlock new potential for artificial intelligence?
Following the two talks, Emily Mackevicius, PhD, Associate Research Scientist at Columbia University’s Zuckerman Institute, and Co-Founder and Director of Basis Research Institute, will moderate a discussion and Q&A with the speakers. Audience questions are welcomed, either submitted during registration or live during the event.
About the experts
Zenna Tavares, PhD, is the first Innovation Scholar at Columbia University's Zuckerman Mind Brain Behavior Institute and an Associate Research Scientist in the Data Science Institute. He also co-founded the Basis Research Institute and serves as its co-director. Dr. Tavares’ research aims to understand how humans reason, that is, how they come to derive knowledge from observing and interacting with the world. To achieve this, he constructs computational and statistical tools to advance causal reasoning, probabilistic programming, and scientific model discovery. Prior to Columbia University, he was at MIT, where he received a PhD in Cognitive Science and Statistics and was a Postdoctoral Researcher in the Computer Science Artificial Intelligence Lab (CSAIL). Dr. Tavares’ work has received significant recognition, including an International Fulbright Science and Technology Award for Outstanding Foreign Students.
Kimberly Stachenfeld, PhD, is a Senior Research Scientist at Google DeepMind in NYC and Affiliate Faculty at the Center for Theoretical Neuroscience at Columbia University. Dr. Stachenfeld’s research spans topics in Neuroscience and Machine Learning. On the Neuroscience side, she studies how animals build (and use) relational models of their world: similarities and differences between different experiences, predictive models of how events are likely to unfold, maps representing relationships between locations and objects in the real world or more abstract concepts. On the Machine Learning side, she works on implementing these cognitive functions in neural networks. She has worked particularly on Graph Neural Networks and Predictive models for simulation of complex physical systems. In 2019, Dr. Stachenfeld was named one of MIT Tech Review’s Innovators under 35 for her work on predictive representations in hippocampus. She completed her PhD at the Princeton Neuroscience Institute in Quantitative and Computational Neuroscience with Matthew Botvinick. Before that, she studied Mathematics (BA) and Chemical & Biological engineering (BS) at Tufts University.
Emily Mackevicius, PhD, is a brain and cognitive scientist who investigates how intelligent behaviors arise in distributed and recurrent systems, with a particular focus on rapid learning, and one-shot memory in birds. Her research bridges multiple levels of abstraction, from neural mechanisms, to cognitive functions, to collaborative group behaviors. Her theoretical work is strongly grounded in experimental practice, including neural and behavioral recordings of birds with extreme memory abilities. Dr. Mackevicius is affiliated with Columbia’s Zuckerman Institute through the Aronov Lab and the Center for Theoretical Neuroscience. In addition, Dr. Mackevicius has co-founded a new research institute, Basis, which develops open-source AI code applied to real-world problems, including understanding collaborative multi-agent behaviors.
This talk is part of the Stavros Niarchos Foundation Brain Insight Lecture series, offered free to the public to enhance understanding of the biology of the mind and the complexity of human behavior. The lectures are hosted by Columbia’s Mortimer B. Zuckerman Mind Brain Behavior Institute and supported by the Stavros Niarchos Foundation.