Ch1 AI intro kho tài liệu bách khoa

What is AI (Artificial
Intelligence)

3/15/2016

1


Topics
• Quotes & Concepts
• Goals & Approaches
• Application Areas
• Framework for AI Systems
• Fundamental Issues for AI Problems

3/15/2016

2


What is AI?
“It is the science and engineering of
making intelligent machines, especially
intelligent computer programs.
It is related to the similar task of using
computers to understand human
intelligence, but AI does not have to
confine itself to methods that are
biologically observable.”
“Intelligence is the computational part of
the ability to achieve goals in the world.”

J. McCarthy
3/15/2016

3


What is AI?
“AI is
the design, study and
construction
of computer programs
that behave intelligently.”
Tom Dean

3/15/2016

4


What is AI?
“AI is the study of
complex information processing problems
that often have their roots in some aspect
of biological information processing.
The goal of the subject is to identify solvable and
interesting information processing problems, and
solve them.”
David Marr (1945-1980)

3/15/2016

5


What is intelligent behavior?
• Perceiving one’s environment
• Acting in complex environments
• Learning and understanding from experience
• Using reasoning to solve problems and
to discover “hidden” knowledge
• Applying knowledge successfully in new situations
• Thinking abstractly, using analogies
• Communicating with others

3/15/2016

6


Goals of AI
• To replicate human intelligence (still distant goal)
• To solve knowledge-intensive tasks
• To make an intelligent connection between
perception and action
• To enhance human-human, human-computer
and computer-computer interaction/communication
– Computer can sense and recognize its users, see and
recognize its environment, respond visually and audibly to
stimuli.
– New paradigms for interacting productively with computers

using speech, vision, natural language, 3D virtual reality,…
3/15/2016

7


Goals of AI
• Engineering Goal
– Develop concepts, theory and practice of building
intelligent machines
– Emphasis on system building

• Science Goal
– Develop concepts, mechanisms and vocabulary to
understand biological intelligent behavior
– Emphasis on understanding intelligent behavior

3/15/2016

8


Approaches to AI
• Choose
– The goals of the computational model
– The basis for evaluating performance of the system

Human-like:
Think: 1) Think like humans
Cognitive science

Act:

3/15/2016

Rational:
2) Think rationally
Laws of thought

3) Act like humans
4) Act rationally
Behaviorist approach
Satisficing methods

9


Goals of and Approaches to AI
• Box 1: Cognitive Science Approach
– Focus on behavior and I/O
– Model reasoning processes
– Computational model should reflect
how results were obtained

• Goal
– Not just to produce human like behavior (box 3)
but to produce a reasoning process similar to the steps
used by humans

3/15/2016

10


Goals of and Approaches to AI
• Box 2: Laws of Thought
– Focus on inference mechanisms that
are provably correct and guarantee an optimal solution

• Goal
– Formalize the reasoning process as a system of logical
rules and procedures for inferencing
– Develop systems of representation to allow inferences to
be performed

3/15/2016

11


Goals of and Approaches to AI
• Box 3: Behaviorist Approach
– Focus on action in the world, intelligent behavior
– Not centered around representation of the world

• Goal
– Develop systems that are human-like

• Turing Test
– Alan Turing 1950
– Can a machine fool an interrogator into

believing that it is a human?
3/15/2016

12


Goals of and Approaches to AI
• Box 4: Satisficing Methods
– Focus on systems that act sufficiently
if not optimally in all situations
– It's OK to have imperfect reasoning
if the job gets done

• Goal
– Develop systems that are rational and sufficient

3/15/2016

13


Goals of and Approaches to AI
• There are two kinds of people in the AI world:
– Classical AI, the symbol-processing approach
• Top down, knowledge to symbol to implementation
• Uses logical operations and
declarative knowledge bases
• Newell and Simon’s physical symbol system

– Sub-symbolic AI, the animat approach

• Bottom up, from signal to symbol
• Design proceeds in steps of the evolutionary ladder
• Intelligence is an emergent property
without requiring centralized models

3/15/2016

14


Some Application Areas of AI
• Game Playing
• Speech Recognition
• Computer Vision
• Expert Systems
• Heuristic Classification
• Natural Language Understanding
• Scheduling and Planning

3/15/2016

15


Some Application Areas of AI
• Game Playing
– <$100 for a machine that significantly uses AI
playstation, game cube, etc.
– Deep Blue beat world champion Gary Kasparov by looking
at hundreds of millions positions per second

• Speech Recognition
– PEGASUS allows users to obtain flight info and make
reservations by speaking over the phone
– Significant advances in speech recognition (and processor
speed) now make systems possible
3/15/2016

16


Some Application Areas of AI
• Computer Vision
– Face recognition programs: banks, casinos, police
– CMU's ALVINN autonomously drove a van from
Washington, D.C. to San Diego, averaging 63 mph day
and night, in all weather conditions
– Handwriting recognition, electronics and manufacturing
inspection, photo-interpretation, baggage inspection,
reverse engineering to automatically construct a 3D
geometric model
– Advances have been made, but humans are
far better at understanding and representing 3D
information
3/15/2016

17


Some Application Areas of AI

• Expert Systems
“knowledge engineer” tries to embody the
knowledge of human experts in a computer program
that is used to carry out some task
– diagnostic systems (MYCIN 1974)
– system configuration (DEC)
– financial decision making

• Heuristic Classification
categorize examples based on guidelines
– credit card purchases (Visa)
3/15/2016

18


Some Application Areas of AI
• Natural Language Understanding
– web page translators (AltaVista)
– Catepillar's truck manuals translated
into 20 languages by machine
– getting a sequence of words and parsing sentences isn’t
enough, understanding requires knowledge of the domain,
so present systems are quite limited

• Scheduling and Planning
– automatic scheduling for manufacturing
– American Airlines rerouting contingency planner
– European Space Agency planning and scheduling of
spacecraft assembly, integration and verification

3/15/2016

19


Natural Language: Chatbots
• ALICE: 2004 Loebner Prize winner
• ELIZA: psychotherapist
• Valerie: CMU Robot Receptionist
• Natural language processing, pattern matching
• Loebner Prize gold medal, awarded annually to
best AI program

3/15/2016

20


A.L.I.C.E. Chatbot (www.alicebot.org)
•
•
•
•
•
•
•
•
•
•
•

•
•
•

Jerry: my name is jerry.
ALICE: Hey jerry.
Jerry: do you know Turing?
ALICE: A brilliant and tragic figure in the history of computer science.
Jerry: who else do you know?
ALICE: Hugh Loebner, Robby Garner, and Steven Spielberg
Jerry: which film do you like?
ALICE: My favorite movie is Teknolust. Have you seen it?
Jerry: no.
ALICE: I would recommend you check it out.
Jerry: what is it about?
ALICE: Science fiction.
Jerry: do you remember my name?
ALICE: Your name is jerry, seeker.

3/15/2016

21


AI Apps Top-10 List
1. Language translation services (Google)
2. News aggregation and summarization (Google)
3. Speech recognition (Nuance)
4. Song recognition (Shazam)
5. Face recognition (Recognizr)

6. Image recognition (Google Goggles)
7. Question answering (Apple Siri, IBM Watson)
8. Chess playing (IBM Deep Blue)
9. 3D scene modeling from images (Microsoft Photosynth)
10. Driverless cars (Google)
3/15/2016

22


Some AI "Grand Challenge" Problems
• Intelligent Agents
• Smart Clothes
• Aids for the Disabled
• Tutors
• Accident-Avoiding Vehicles
• Self-Organizing Systems
• Translating Telephone Conversations
• Extracting and representing information from lots of data
– Neural networks, hidden Markov models,
– Bayesian networks, heuristic search, logic, …
3/15/2016

23


A Framework for Building AI Systems
• Perception
• Reasoning
• Action

3/15/2016

24


Perception
• Biological systems experience the world
through their senses
• What perceptions would be needed by:
– autonomous vehicle? camera images and range data
– medical diagnosis system? symptoms and test results

• Includes areas of
–
–
–
–

vision
speech processing
natural language processing
signal processing, eg: market data, acoustic data

3/15/2016

25