1 / 15

2008. 4. 14 B.-W. Ku

The St. Thomas Common Sense Symposium: Designing Architectures for Human-Level Intelligence Marvin Minsky, Push Singh, Aaron Sloman, AI Magazine 2004. 2008. 4. 14 B.-W. Ku. A Symposium in St. Thomas (2002). To discuss a project–to develop new architectural schemes that can

zander
Télécharger la présentation

2008. 4. 14 B.-W. Ku

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The St. Thomas Common Sense Symposium:Designing Architectures for Human-Level IntelligenceMarvin Minsky, Push Singh, Aaron Sloman, AI Magazine 2004 2008. 4. 14 B.-W. Ku

  2. A Symposium in St. Thomas (2002) • To discuss a project–to develop new architectural schemes that can bridge between different strategies and representations. • Participants: Larry Birnbaum (Northwestern Univ.) Ken Forbus (Northwestern Univ.) Ben Kuipers (Univ. of Texas at Austin) Douglas Lenat (Cycorp) Henry Lieberman (MIT) Henry Minsky (Laszlo Systems) Marvin Minsky (MIT) Erik Mueller (IBM Watson Research Center) Srini Narayanan (U. C. Berkeley) Ashwin Ram (Georgia Institute of Technology) Doug Riecken (IBM Watson Research Center) Roger Schank (Carnegie Mellon Univ.) Mary Shepard (Cycorp) Push Singh (MIT) Jeffrey Mark Siskind (Purdue Univ.) Aaron Sloman (Univ. of Birmingham) Oliver Steele (Laszlo Systems) Linda Stone (independent consultant) Vernor Vinge (San Diego State Univ.) Michael Witbrock (Cycorp) (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  3. The Need for Synthesis in Modern AI • Most researchers in recent years have retreated from the ambitious aim of building a machine that has common sense. • Instead, each developed some special technique that could deal with some class of problem well, but does poorly at almost everything else. • To build a human-level AI, we must develop ways to combine the advantages of multiple methods. (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  4. Organizing the Diversity of AI Methods • We need a theory that helps us to map the types of problems onto the types of solutions. • Minsky’s suggestion: The Causal Diversity Matrix (Fig. 1) (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/ Figure 1

  5. Figure 2 Returning to the Blocks World • A possible target problem domain for a commonsense architecture project • Multiple realms • Spatial, physical, bodily and visual reasoning • Psychological, social, reflective, conversational and educational reasoning • A model world • Physically realistic • Social problems • Simulated beings participate in scenarios including: • Jointly building structures • Competing to solve puzzles • Teaching each other skills • Verbally reflecting on their own successes and failures • Video game community – programmable virtual world (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  6. Large-Scale Architectures for Human-Level Intelligence (1) • How could a meta-theory (as in Fig. 1) of AI techniques be used by an AI architecture? (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  7. Large-Scale Architectures for Human-Level Intelligence (2) • Minsky’s emotion machine architecture (Fig. 4) • Reflective thinking Figure 4 (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  8. Large-Scale Architectures for Human-Level Intelligence (3) • Select some “ways-to-think” • Adaptation or switching • Active in parallel Figure 6 (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/ Figure 7

  9. Large-Scale Architectures for Human-Level Intelligence (4) • Sloman’s H-CogAff model (Fig. 5) • Cognition and Affect project has explored models for human minds. (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/ Figure 5

  10. Educating the Architecture (1) • How to supply the architecture with a broad range of commonsense knowledge? • Learning • Start out with too little knowledge would not likely to achieve enough versatility. • Minsky’s complaint about “baby machines” • We don’t know how to do it yet. • Such approaches have all failed to make much progress (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  11. Educating the Architecture (2) • General-purpose commonsense knowledge resources • Cyc system • A great many way to represent commonsense knowledge • Over a million (< adult-level: 100 million) commonsense facts and rules • Move to a distributed knowledge acquisition approach • Volunteer teachers around the world needed • Development of friendly interfaces • Criticism: • Not adequately based on cognitive science • Minsky • Need to augment each existing item with procedural and heuristic knowledge, such as descriptions of: target problems; ways of thinking; known arguments for and against using it; ways to adapt it to a new context (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  12. An Important Application • To receive substantial support for such a project. • A personalized teaching machine • It adapt itself to someone’s particular circumstances, difficulties, and needs. • It carry out a conversation with you. • You could discuss with it various subjects. (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  13. Final Consensus • An architecture that can support many different techniques • A model world • A personalized teaching machine (many, but not all, agreed) (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  14. A Collaborative Project • It’s feasible when the components can be reasonably disassociated. • Some technical steps • A virtual model world • Miniscenarios • Protocols • A catalog of ways-to-think • Self-reflections • Commonsense knowledge base • “Intention-based” programming language (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/

  15. Summary (C) 2008 SNU Biointelligence Lab  http://bi.snu.ac.kr/ Figure 8

More Related