Weak AI: Can Machines Act Intelligently

1
Weak AI Can Machines Act Intelligently?

• Some things they can do
• Computer vision face recognition from a large
  set
• Robotics autonomous (mostly) car
• Natural language processing simple machine
  translation
• Spoken language systems 1000 word continuous
  speech
• Learning text categorization into 1000 topics
• Games Grand Master level in chess (world
  champion), etc.
• Some ways they do it
• Search
• Knowledge Representation
• Machine Learning
• So -- yes or no?

2
Strong AI Can Machines Really Think?

• Argument from consciousness, Chinese Room
• Computer programs are formal, syntactic entities
• Minds have mental contents, or semantics
• Syntax is not by itself sufficient for semantics
• Brains cause minds.
• What do we mean by "really think"?

3
Some Ethical Issues in AI

• The usual automation issues
• loss of jobs
• dependence on technology
• loss of motivation
• Privacy issues
• Loss of accountability
• Danger to humanity (Asimov's Three Laws, Vinge's
  Singularity)
• Do AIs have rights?

4
What Next?

• 2003 IJCAI Invited Speakers
• Computer Vision AI or Non-AI Problem (Kanade,
  CMU)
• The Past, Present and Future of Web Information
  Retrieval (Henzinger, Google)
• Deploying Information Agents on the Web
  (Knoblock, S. Cal)
• Web Intelligence (WI) A New Paradigm for
  Developing the Wisdom Web and Social Network
  Intelligence (Liu, Hong Kong Baptist University)
• Intelligent Systems in Travel and Tourism
  (Werthner, University of Trento, Italy)
• Optimality of Collective Choice in Social Insects
  and Social Robots (Deneubourg, University Libre
  du Bruxelles)
• Self-reconfiguring Robots Challenges and
  Successes (Rus, Dartmouth)
• And papers for 2004

5
AAAI 2004 Some Workshops

• Adaptive Text Mining and Extraction
• Agent Organizations, Coalitions and Teams in
  Adaptive MultiAgent Systems
• Challenges in Game AI
• Semantic Web Personalization
• Sensor Networks, Anchoring Symbols to Sensor Data
  
• Intelligent Agent Architectures
• Collective Mind Architectures for Fleets of
  Equipment that Learn from their Experience

6
2003 IAAI Deployed Applications Papers

• Criterion Online Essay Evaluation An Application
  for Automated Evaluation of Student Essays
• Mobile Intelligence for Door-to-Door Sales
  Support System
• The NASD Securities Observation, New Analysis,
  and Regulation System (SONAR)
• TPO A System for Scheduling and Managing Train
  Crew in Norway

7
2003 IAAI Emerging Applications Papers

• Applying Reinforcement Learning to Packet
  Scheduling in Routers
• The Analogical Thesaurus
• Broadcast News Understanding and Navigation
• Qualitative Spatial Reasoning about Sketch Maps
• Infrastructure Components for Large-Scale
  Information Extraction Systems
• Building Agents for the Customer Service Front
• Searching for Hidden Messages Automatic
  Detection of Steganography
• A Probabilistic Vehicle Diagnostic System Using
  Multiple Models
• A Knowledge Acquisition Tool for Course of Action
  Analysis
• Teachable Agents A Learning by Teaching
  Environment for Science Domains
• Secure Mobile Agents on Ad Hoc Wireless Networks
• Say Cheese! Experiences with a Robot
  Photographer
• Transparent Grid Computing A Knowledge-Based
  Approach
• LAW A Workbench for Approximate Pattern Matching
  in Relational Data
• Guided Conversations about Leadership

8
Where Will We Be in 25 Years?
2004 ?? 2029 ??
9
And what if we get there?