Transcript RAIR Lab Visit, Ken Gertz 010604

RAIR Lab Visit, Ken Gertz
010604
(draft v3; final)
Rensselaer AI & Reasoning (RAIR) Lab Personnel:
Faculty: Selmer Bringsjord,
Konstantine Arkoudas, Marc Destefano, Jim Fahey,
Bram van Heuveln, Ron Sun, Yingrui Yang, Michael Zenzen
GRAs: Paul Bello, Bettina Schimanski
URAs: Andrew Shilliday, Josh Taylor, Owen Kellett
RAIR Lab Visit, Ken Gertz
010604
(draft v3; final)
Rensselaer AI & Reasoning (RAIR) Lab Personnel:
Faculty: Selmer Bringsjord,
Konstantine Arkoudas, Marc Destefano, Jim Fahey,
Bram van Heuveln, Ron Sun, Yingrui Yang, Michael Zenzen
GRAs: Paul Bello, Bettina Schimanski
URAs: Andrew Shilliday, Josh Taylor, Owen Kellett
The Problem
(Perhaps not Coincidentally:
Institutions Without “Cognitive Power”)
Dollars in Thousands - 1999
Total R&D Expenditures Compared
1,000,000
900,000
800,000
700,000
600,000
500,000
400,000
300,000
200,000
100,000
0
ch
e
aT
G
Total 99
CM
U
I
RP
JH
M
IT
S
d
or
f
n
ta
Institutions
nn
e
P
W
No
n
er
t
es
a
an
i
d
In
Looking at Trends Paints the Same Picture
2001
Total R&D Expenditures
1,000,000
How?
Dollars in Thousands
900,000
800,000
700,000
RPI
600,000
MIT
500,000
Johns Hopkins
400,000
Georgia Tech
300,000
200,000
100,000
0
1992 1993 1994 1995 1996 1997 1998 1999
Years
2010
What is Cognitive Science?
• Cognitive Science is the attempt to understand, simulate, and replicate
cognitive systems or intelligent agents (which are embodied in
minds/brains/machines), in information processing and (frequently)
representational schemes.
• Cognitive Science can be understood as the attempt to harness IT in order to
understand, simulate, and replicate the mind/brain.
• Cognitive Science includes relevant parts of:
–
–
–
–
–
–
–
Artificial Intelligence
Psychology
Philosophy/Logic
Educational Technology
Linguistics
Neuroscience
Anthropology
= our disciplines
The Cognitive Science Department’s
Four Foci…
Perception and Action
High-level
Low-level
Perception
Environment
subdeclarative computation
Cognitive System
Action
Computational Cognitive Modeling
Low-level
High-level
Perception
Cognitive System
Environment
Perception
& Action
subdeclarative computation
Long Term
Memory
Clarion, ACT-R
Short Term
Memory
Action
Human & Machine Reasoning
Low-level
High-level
Perception
Perception
& Action
Cognitive System
Environment
Long Term
Memory
subdeclarative computation
Short Term
Memory
ACT-R
Semantic
Reasoning
Mental Metalogic
Syntactic
Reasoning
Action
Cognitive Engineeering
Low-level
High-level
Perception
Perception
& Action
Cognitive System
Environment
Long Term
Memory
subdeclarative computation
Short Term
Memory
ACT-R
Semantic
Reasoning
Mental Metalogic
Syntactic
Reasoning
Action
The Rensselaer AI & Reasoning Lab
(The RAIR Lab)
Busy Beaver Project;
Intelligent
Tutoring Systems
(mathematical logic)
Over
$1million
internal
seeding
Slate (Intelligence
Analysis)
Item generation
(theorem proving-based
generation)
synthetic characters/psychological time
The Paradox of Human vs.
Machine Reasoning
• On the one hand, machine reasoners are
getting faster, and can do some impressive
things.
• On the other hand, as Herb Simon, one of
the grandfathers of AI, admitted before he
died, machine reasoning is absolutely
nowhere when stacked against first-rate
human reasoning.
Response:
Next-Generation AI
publication-wise, lots:
e.g.,
Ron Sun has joined CSD!
Yingrui Yang is prolific!
see CD
Proof using Hybrid
Reasoning from
Mental MetaLogic
(Proof Construction in
Hyperproof)
Applications (interconnected)
• Defense
– intelligence analysis (ARDA)
– cognitive architectures (Clarion, ACT-R, Soar)
– “3rd generation” wargaming (w/ AFRL)
• Frontier-Breaking Logic & AI/Computer Science
–
–
–
–
Athena
MARMML/CHOGIC
EGs, ATP based in  and 
The  Cracking (“Busy Beaver”) Project (NSF)
• AI & Interactive Digital Entertainment
– State-of-the-art in teaching game development
– Including advanced synthetic characters
– Co-Locational Digital Entertainment (CLIDE)?
• AI & Education/Edutainment
– intelligent tutoring systems (ITS)
– Including, again, advanced synthetic characters
• Cognitive Prostheses (exploratory only at this stage)
Some Upcoming Props
• NSF -> Ed -> K-12 -> IMD (Simlog;
Chogic)
• NSF -> Ed -> CCLI-EMD (ITS; in-person
meet in DC before submission)
• Retreat to seed $ prop, DARPA
• Seed $ AFRL (tomorrow to Rome)
• NSF CyberTrust?
Demo Sequence (blue ones on for 1.6.04)
•
•
Selmer gives overview start @ 205p; done @ 215 (all slots include reasonable Q&A)
Busy Beaver Competition (SB for Owen Kellett) 215-220
–
•
•
•
•
•
•
•
NIMD, Slate & a simple scenario (Josh Taylor) 220-235
Slate & empirical investigation of reasoning using cave & recording of subjects tackling
micro-scenario (Jason Wodicka)
Advanced Synthetic Characters (Selmer et al.) ppt on CD
Clarion demo (Ron Sun) Ron is in China!
EG-based machine reasoning (Bram van Heuveln)
 & PERI (Bettina Schimanski) 235-250
Athena & railroad example (Kostas Arkoudas) 250-310
–
–
•
includes best two games created by RPI students
uses Paul’s slides, and alludes to exploratory collaboration with Rome AFRL
includes quick mention of commercial wargames/strategy-based conflict
Introductory Cognitive Robotics (Marc Destefano) 325-335
–
•
•
include run showing soundness of UNIX OS, and brief explanation
tie to CyberTrust
Digital Entertainment (including wargaming) (Marc Destefano (representing AFRL’s
Paul Bello as well)) 310-325
–
–
–
•
http://www.cs.rpi.edu/~kelleo/busybeaver
use video I just used in DC?
Superteaching (Marc Destefano) 335-345
Discussion, Q&A etc. last 15 minutes