Panel Discussion I: Brainstorm on Language, Embodiment and the Critical minass of Intelligence Moderator: Alexei Samsonovich Panelists: Kenneth De Jong,

Slides:



Advertisements
Similar presentations
Cognitive Systems, ICANN panel, Q1 What is machine intelligence, as beyond pattern matching, classification and prediction. What is machine intelligence,
Advertisements

Outline Administrative issues Course overview What are Intelligent Systems? A brief history State of the art Intelligent agents.
ARCHITECTURES FOR ARTIFICIAL INTELLIGENCE SYSTEMS
Breakout session B questions. Research directions/areas Multi-modal perception cognition and interaction Learning, adaptation and imitation Design and.
SPECIFYING MODALITIES IN THE MGLAIR ARCHITECTURE Stuart C. Shapiro and Jonathan P. Bona Department of Computer Science and Engineering And Center for Cognitive.
Computation and representation Joe Lau. Overview of lecture What is computation? Brief history Computational explanations in cognitive science Levels.
The GLAIR Architecture for Cognitive Robotics Stuart C. Shapiro Department of Computer Science & Engineering and Center for Cognitive Science.
The GLAIR Cognitive Architecture and Prospects for Consciousness Stuart C. Shapiro Department of Computer Science & Engineering and Center.
The GLAIR Cognitive Architecture Stuart C. Shapiro and Jonathan P. Bona Department of Computer Science & Engineering Center for Cognitive Science.
The GLAIR Architecture for Cognitive Robots Stuart C. Shapiro Department of Computer Science & Engineering and Center for Cognitive Science.
Knowledge Representation for Self-Aware Computer Systems Stuart C. Shapiro Department of Computer Science and Engineering, and Center for Cognitive.
NCTM’s Focus in High School Mathematics: Reasoning and Sense Making.
Knowledge Acquisitioning. Definition The transfer and transformation of potential problem solving expertise from some knowledge source to a program.
Cassie as a Self-Aware SNePS/GLAIR Agent Stuart C. Shapiro Department of Computer Science and Engineering, and Center for Cognitive Science.
Intelligence without Reason
S.C. Shapiro Symbol-Anchoring in Cassie Stuart C. Shapiro and Haythem O. Ismail Department of Computer Science and Engineering and Center for.
S.C. Shapiro Symbol Anchoring in a Grounded Layered Architecture with Integrated Reasoning Stuart C. Shapiro Department of Computer Science.
Robots at Work Dr Gerard McKee Active Robotics Laboratory School of Systems Engineering The University of Reading, UK
Biointelligence Laboratory School of Computer Science and Engineering Seoul National University Cognitive Robots © 2014, SNU CSE Biointelligence Lab.,
Vedrana Vidulin Jožef Stefan Institute, Ljubljana, Slovenia
THEORIES OF MIND: AN INTRODUCTION TO COGNITIVE SCIENCE Jay Friedenberg and Gordon Silverman.
Learner Diversity and Classroom Learning. Classroom Management not a set of discipline and control strategies to make students to work and listen to teacher.
Self-Organized Recurrent Neural Learning for Language Processing April 1, March 31, 2012 State from June 2009.
Artificial Intelligence CIS 479/579 Bruce R. Maxim UM-Dearborn.
Chapter 14: Artificial Intelligence Invitation to Computer Science, C++ Version, Third Edition.
Steps Toward an AGI Roadmap Włodek Duch ( Google: W. Duch) AGI, Memphis, 1-2 March 2007 Roadmaps: A Ten Year Roadmap to Machines with Common Sense (Push.
INTEGRATED SYSTEMS 1205 Technology Education A Curriculum Review Sabine Schnepf-Comeau July 19, 2011 ED 4752.
An Introduction to Artificial Intelligence and Knowledge Engineering N. Kasabov, Foundations of Neural Networks, Fuzzy Systems, and Knowledge Engineering,
What is AI:-  Ai is the science of making machine do things that would requires intelligence.  Computers with the ability to mimic or duplicate the.
 The most intelligent device - “Human Brain”.  The machine that revolutionized the whole world – “computer”.  Inefficiencies of the computer has lead.
Towards Cognitive Robotics Biointelligence Laboratory School of Computer Science and Engineering Seoul National University Christian.
SEMINAR REPORT ON K.SWATHI. INTRODUCTION Any automatically operated machine that functions in human like manner Any automatically operated machine that.
Putting Research to Work in K-8 Science Classrooms Ready, Set, SCIENCE.
Synthetic Cognitive Agent Situational Awareness Components Sanford T. Freedman and Julie A. Adams Department of Electrical Engineering and Computer Science.
+ BICA Challenge Panel Discussion Session 1 BICA VideoPanels * May 5, 2011 A virtual extension of BICA 2011 Panelists: Andrea Stocco University of Washington.
Theories of First Language Acquisition
Artificial Intelligence By Michelle Witcofsky And Evan Flanagan.
EDN:204– Learning Process 30th August, 2010 B.Ed II(S) Sci Topics: Cognitive views of Learning.
COGNITIVE SEMANTICS: INTRODUCTION DANA RETOVÁ CSCTR2010 – Session 1.
University of Windsor School of Computer Science Topics in Artificial Intelligence Fall 2008 Sept 11, 2008.
1 The main topics in AI Artificial intelligence can be considered under a number of headings: –Search (includes Game Playing). –Representing Knowledge.
Cognitive Systems Foresight Language and Speech. Cognitive Systems Foresight Language and Speech How does the human system organise itself, as a neuro-biological.
Technical Goals for the BICA Community Mark R. Waser
CS 4620 Intelligent Systems. What we want to do today Course introductions Make sure you know the schedule for the next three weeks.
RULES Patty Nordstrom Hien Nguyen. "Cognitive Skills are Realized by Production Rules"
Vedrana Vidulin Jožef Stefan Institute, Ljubljana, Slovenia
Miss. Mona AL-Kahtani.  Basic assumption:  Language acquisition is one example of the human child’s remarkable ability to learn from experience and.
Artificial Intelligence: Research and Collaborative Possibilities a presentation by: Dr. Ernest L. McDuffie, Assistant Professor Department of Computer.
Bridges To Computing General Information: This document was created for use in the "Bridges to Computing" project of Brooklyn College. You are invited.
Slide no 1 Cognitive Systems in FP6 scope and focus Colette Maloney DG Information Society.
Introduction: What is AI? CMSC Introduction to Artificial Intelligence January 3, 2002.
INTRODUCTION TO COGNITIVE SCIENCE NURSING INFORMATICS CHAPTER 3 1.
Introduction: What is AI? CMSC Introduction to Artificial Intelligence January 7, 2003.
ASSEMBLY AND DISASSEMBLY: AN OVERVIEW AND FRAMEWORK FOR COOPERATION REQUIREMENT PLANNING WITH CONFLICT RESOLUTION in Journal of Intelligent and Robotic.
Introducing Critical and Creative Thinking. Agenda The importance of Critical and Creative Thinking What is in the curriculum? Questions Planning for.
HIERARCHICAL TEMPORAL MEMORY WHY CANT COMPUTERS BE MORE LIKE THE BRAIN?
Christina Pelletier Columbus State University
What is cognitive psychology?
Learning Fast and Slow John E. Laird
Machine Learning overview Chapter 18, 21
CHAPTER 1 Introduction BIC 3337 EXPERT SYSTEM.
Overview of Year 1 Progress Angelo Cangelosi & ITALK team
Bioagents and Biorobots David Kadleček, Michal Petrus, Pavel Nahodil
Interdisciplinary research on language & speech
Course Instructor: knza ch
Multimodal Human-Computer Interaction New Interaction Techniques 22. 1
MGLAIR Modal Grounded Layered Architecture with Integrated Reasoning
Toward a Great Class Project: Discussion of Stoianov & Zorzi’s Numerosity Model Psych 209 – 2019 Feb 14, 2019.
Educational Technology Lab, National Kapodistrian
Technology of Data Glove
Presentation transcript:

Panel Discussion I: Brainstorm on Language, Embodiment and the Critical minass of Intelligence Moderator: Alexei Samsonovich Panelists: Kenneth De Jong, Dennis Perzanowski, Stuart Shapiro, Ralph Weischedel, Antonio Chella, Wei Chen, Roberto Pirrone, Christian Lebiere

Objective: Agenda: Samsonovich3 min Language acquisition by children Perzanowski5 min Goertzel5 min Weischedel5 min Bridging the gaps Lebiere5 min Chella5 min Shapiro5 min Identifying a critical mass Chen5 min Pirrone5 min De Jong5 min Response from the audience general discussion12 min To identify a sufficient language component of a super-critical mass of intelligence and embodiment that would enable a bootstrapped cognitive growth of an artifact up to a human level of general intelligence Introduction

language acquisition by children

Questions to Dennis Perzanowski: The work of Deb Roy directed at learning how children acquire language implies understanding the cognitive steps and stages of develoopment, and then emulating and embodying them on robotic platforms. – What should we borrow from these studies of the acquisition of language by young children? – E.g., how the agent will develop an understanding of concepts associated with linguistic constructs?

Questions to Ralph Weischedel: If language is the key to the human mind, then understanding the process of language acquisition could be the key to developing human-level intelligence. Specific questions include: – What should we borrow from studies of the acquisition of language by young children? – E.g., how the agent will develop an understanding of concepts associated with linguistic constructs? See next 3 slides prepared by Ralph Weischedel

6 What should we borrow from studies of the acquisition of language by young children? Ralph Weischedel November 7, 2008

7 An Interesting Question, Given History 70s-80s –Computational Linguistics –Hand-crafted rules –Reason from a handful of examples/counter-examples 90s –Natural Language Processing –Statistical learning approaches –Empirical evaluation on blind test data 2000s –90s + –Semi-supervised learning from WWW 2010s? –Inspiration of child language acquisition?

8 What should we Borrow from Child Language Acquisition? Positively Inspiring Leaning paradigm –Training data Image/video + description Touch (+ description) –Output capability Brief utterances Actions (move, touch, push) Simultaneous generalization in multiple spaces –Image space Forming object/activity class –Speech signal –Lexicon What might be innate (and built in for a system) Children interact with –Care-givers –Environment Not Biologically Bound Machines have vast amount of information (text, images and, video) online –A basis for unsupervised learning

One reply from the audience

Bridging the gaps

Questions to Christian Lebiere: The relation between language and the cognitive architecture, including its basic components, memory systems, kinds of representations, and principles of information processing. Specific questions include: – Is language processing an underlying capacity for other faculties like symbolic processing, or vice versa? – What kinds of architectural mechanisms should be involved in language processing?

Questions to Antonio Chella: The kinds of representation and information processing needed for the critical mass. Specific questions include: – What is the missing link between the following two levels of representations? (i) The higher symbolic level that underlies language, reasoning, attention control, planning and decision making. (ii) The lower subsymbolic level associated with neural network dynamics and signal processing. – Is this missing link a missing component of the critical mass? – How the conceptual spaces proposed by Gärdenfors (2000) may allow us to bridge this gap?

Questions to Stuart Shapiro: An example of a potential critical mass: GLAIR architecture that uses symbols in the Knowledge Layer as "pivots" that each may be aligned with one or more bodily modalities. Thus, one can identify (in NL) and manipulate what one sees, or one may find in the environment what one hears about (via NL). More work is still needed on self-awareness, and on identifying the basic cognitive building blocks that underlie all languages and cultures, and that therefore allow an intelligent agent to learn any language and culture it is brought up in. Specific questions include: – What are the general critical mass requirements for the embodiment and the interface, including environment, domain, perception and action capabilities? – What are the general critical mass requirements for the communication capabilities, including initial natural language capabilities? – What are the general critical mass requirements for the "innate" (pre-programmed) general knowledge? See next 3 slides prepared by Stuart Shapiro

KL (SNePS) PMLa PMLb PMLc SAL Mind Body Independent of lower-body implementation Hearing Vision Motion Speech WORLDWORLD I/P s o c k e t s MGLAIR Agent Architecture Dependent on lower-body implementation Proprioception

Alignment KL Body (PML/SAL) World SNePS term PML structure Object/PhenomenonAction

Jackendoff Version Ray Jackendoff, Foundations of Language, Oxford University Press, 2002, Fig. 11.1, p (Independent work)

One reply from the audience

Identifying a (super)critical mass

Questions to Wei Chen: The critical mass of linguistic, cognitive and interface capabilities that will enable cognitive growth of an artifact. Specific questions include: – What are the general critical mass requirements for the memory systems, kinds of representations and principles of information processing? – What are the general critical mass requirements for the communication capabilities, including initial natural language capabilities? – What are the general critical mass requirements for the "innate" (pre-programmed) general knowledge?

Questions to Roberto Pirrone: The connection between language in AI and educational sciences, specifically, language understanding techniques used in education. Specific questions include: – What is the applicability of Cognitive Linguistics techniques to the task of understanding language in educational practice? – What is the need for a language understanding component in a framework for student tutoring, and how this component should be implemented?

Questions to Kenneth De Jong: The notion of a critical mass apparently involves higher cognitive abilities like metacognition and the sense of self. – Can we identify a (super)critical mass in terms of higher cognitive abilities? – What is the connection with other aspects, e.g., linguistic and evolutionary requirements?

General discussion and conclusions "Critical mass" can be understood as a minimal set of intelligent capabilities and features of embodiment, such that, when integrated in an artifact, together will enable its cognitive growth up to a human level via teaching, social interactions and guided exploration of knowledge resources and environments. Then what, in your view, are its critical components, and how close we are to creating them? What can you say regarding "the ladder", i.e., the curriculum for artifacts understood as the sequence of tasks, paradigms and intermediate learning goals that will scaffold their rapid cognitive growth up to the human level? Can you imagine a successful cognitive growth scenario: e.g., as initially discussed within BICA program, from a simple egg hunt to a general human level of intelligence? Reproducing the cognitive growth phenomenon in a lab is a fundamental scientific problem. The expected phenomenon, however, has a known biological counterpart: child development. What should we borrow from biological, cognitive, educational, social, etc. sciences in order to ensure success? How can we imagine a roadmap to human-level-intelligent artifacts? What would be a sequence of waypoints and a realistic timetable for reaching them? What should be the cost of the project?