Intelligence: From the Individual to the Collective

The Science Board Symposium is an annual event that brings together members of SFI’s scientific community to reflect on the previous year and look ahead and collaborate on new ideas. After the day and a half symposium, the Science Board will meet separately to discuss general directions in science & complexity science and SFI’s response to both. Typically an event reserved for SFI Researchers, we are delighted to announce that this year, ACtioN members are invited to participate in the general session virtually.

 

ABSTRACT

The year 1815 was an important milestone in the history of mechanized thought. George Boole and Ada Lovelace were both born this year and each contributed to a shift in the way we think about thought: how our most lofty abstract concepts might be underpinned by simple repetitions of “the symbolic language of a calculus.”

Over the last 200 years the outsourcing of ‘intelligent’ thought has proceeded apace -- from the energetically demanding 1946 ENIAC with its 17000 vacuum tubes that dimmed the lights of Philadelphia when activated, to the 2020, 96 layered, 175 billion parameter model of OpenAI’s GPT-3 language model.

In this Science Board meeting researchers shall discuss the uses and abuses of mechanized intelligence, AI, or ML for complexity science research. Both ML and Complexity science have targeted the same domain of largely irreducible complex phenomena - physiology, ecology, society, etc. Both fields use simple, comprehensible mathematics to generate models and theory, but the resulting models and theory they produce to apply to complex phenomena are effectively incomparable. The first is coarse-grained, highly approximate for a broad range of phenomena, and comprehensible. The second is fine-grained, highly predictive for a narrow range of phenomena, and incomprehensible.

Our research community will contemplate the following questions throughout the symposium:

1. 1. How might complexity science integrate and or cooperate with ML/AI approaches to complexity?
2. 2. How important are either prediction or understanding for complexity science?
3. 3. How might SFI engage in this space to best advantage?
4. 4. What ongoing and innovative projects and researchers should we connect to SFI to most effectively explore this emerging domain?

 

AGENDA

FRIDAY, SEPTEMBER 10

9:00 AM to 10:15 AM	The Intelligence of Murray Gell-Mann Panelists: Jim Hartle (UC Santa Barbara & SFI), George Johnson (Author/Journalist), Seth Lloyd (MIT), Lisa Randall (Harvard University), Geoffrey West (Santa Fe Institute). Moderator: David Krakauer (Santa Fe Institute).
10:15 AM to 10:30 AM	Welcome & Intro Dan Schrag (Harvard University & SFI) and Melanie Mitchell (Santa Fe Institute)
10:30 AM to 11:30 AM	Learning: Babies, Bodies, Brains, and Machines Linda Smith (Indiana University)
11:30 AM to 12:30 PM	Towards Collective A.I. Radhika Nagpal (Harvard University)
12:30 PM	Adjourn

 

SATURDAY, SEPTEMBER 11

9:00 AM to 10:00 AM	The Interplay Between AI and Brain Imaging Research Tom Mitchell (Carnegie Mellon University & SFI)
10:00 AM	Adjourn

 

This event is supported, in part, by the National Science Foundation Grant Number 1940594, under the Universality and Accident in Complexity: The Odyssean Quest of Murray Gell-Mann project. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.