CHAPTE

R 15
Artificial
Intelligence
WHERE DO WE GO
FROM HERE?
Sign up for presentation
Reading Quiz 15 Due 11/24
Thanksgiving – Week 15
Presentations Week 16
Final Exam
 70 questions
 120 minutes
 100 points
LEARNING OBJECTIVES
Descri Describe the two types of artificial
be intelligence

Explain the pros and cons of various

Explain knowledge representation methods
Explain the parts of a simple neural
Explain network, how it works, and how it can
incorporate machine learning
Descri Describe how intelligent state–space
be search algorithms work
Give examples of possible usage for each
Give of the following: swarm intelligence,
intelligent agents, and expert systems
LEARNING OBJECTIVES

Explain what a robot is, and list

some tasks for which robots are
currently suited

Explain what a drone is, and list

some tasks drones can perform
INTROD
UCTION
Artificial Intelligence
(AI): creating computer
systems that exhibit
aspects of intelligence

The Turing Test

 Human judge
questions two hidden
entities
 One entity is a person
 One entity is a
computer
 If judge cannot
distinguish computer
from person, then
computer is
intelligent!
AI VIDEO
AI ON PHONE
A DIVISION OF LABOR

Computational Example: managing

tasks a payroll

Example:
Recognition tasks recognizing an
individual’s face

Example: planning
Reasoning tasks your major in college
 Computational tasks
 Typically have algorithmic
solutions
 Computers perform faster
than humans
A  Computers perform more
DIVISIO accurately than humans

N OF Recognition tasks
 Process massive amounts of
LABOR sensory information
 Access massive amounts of
past experience
 Require approximation
 Humans perform much better
than computers
 Reasoning tasks
 Formal reasoning can be
automated to some extent
 Problems become intractable
quickly
 Common-sense reasoning
A  Requires great experience and
DIVISIO knowledge

N OF
LABOR
KNOWLEDGE
REPRESENTATION
How can we represent
knowledge for the computer?
 Natural language
 Use requires understanding of the meanings of
words and combinations of words
 “Spot is a brown dog” or “Every dog has four
legs”
 Formal language: the language of
formal logic
 “Spot is a brown dog” becomes dog(Spot) AND
brown(Spot)
 “Every dog has four legs” becomes “For every
x, if x is a dog then x has four legs” (∀x) dog(x)
-> four-legged(x)
 Requirements of
a representation
 Adequacy: must
KNOWLED capture all relevant
GE information
 Efficiency: avoid
REPRESEN redundant information
 Extendability: easy
TATION to add new knowledge
 Appropriate: easy
to use for particular
purpose
 Some AI work attempts to mimic the brain
Humans have 86 billion (1012) neurons
A neuron receives electrical stimuli from
other neurons through dendrites
A neuron sends electrical stimuli through
its axon

RECOG Signals pass through gaps, synapses

Some synapses cause increased
NITION activation; others inhibit activation

TASKS Neurons are like very simple

computational devices
RECOGNITI
ON TASKS

Artificial neural networks mimic

the connectionist approach

A threshold for generating

output
Individual artificial An activation level
“neurons” have: Incoming weighted edges
Outgoing weighted edges
REASONING TASKS
INTELLIGENT
SEARCHING
Decision tree represents possible next items
for which to search
Linear search and binary search assume:
 Data is organized linearly
 Exact match is required

What if we relax the requirements?

 What if data is not linear?
 What if an approximate match is okay?
REASONING TASKS
INTELLIGENT
SEARCHING
REASONING TASKS
INTELLIGENT
SEARCHING
State–space graph
 Each node is a state of our problem
 A node connects to another if that state can be directly
generated by the node
 Examples: tic-tac-toe, eight-puzzle, maze-solving
 Each node has many children
 May be many paths to a goal

State–space search: seeks a path from start

state to goal state
REASONING TASKS
INTELLIGENT
SEARCHING
Searching for a path to a goal
 Brute force: trace all branches of decision
tree
 Too slow
 Heuristics: use educated guess to guide
which branches to search
REASONING TASKS
SWARM
INTELLIGENCE
Swarm intelligence model
 Model communities of simple agents, e.g.,
ants, termites, etc.

Ant colonies
 Individuals exhibit simple behaviors
 Colonies accomplish great things
 Finding the shortest path to food
 Constructing nests

Ant colony optimization: route-

finding using simulated ants
REASONING TASKS
INTELLIGENT AGENTS
Intelligent agent works with human
user
 Learns user’s preferences and takes actions on
user’s behalf

Current examples
 Personalized web search (push technology)
 E-commerce site that tailors suggestions to your
interests (recommendation software)

Future applications
 Personal travel planner: buys tickets for user
 Office manager: screens calls, arranges meetings
Expert system: mimics reasoning in some specific
domain
Knowledge base: knowledge about a domain
Inference engine: rules for reasoning with knowledge
Often use formal language to represent knowledge and
rules for inference
Employ deductive reasoning, e.g., modus ponens

REASONING
TASKS EXPERT
SYSTEMS
Expert system reasoning
Forward chaining
 Start with assertions ► look for rules to deduce new assertions
 Given assertion A and rule “if A then B” ► deduce B

Backward chaining
 Start with a query ► look for rules that could deduce query
 Given question “Is B true?” and rule “if A then B” ► try to
determine “Is A true?”

REASONING
TASKS EXPERT
SYSTEMS
Explanation facility
 Users can see explanation based on the reasoning
chain
Knowledge engineering
 Human system builders must spend time with
experts
 Listing and codifying the expert knowledge

REASONING
TASKS EXPERT
SYSTEMS
REASONING
TASKS THE
GAMES WE PLAY
Board games
 Many programs use forms
of state-space search
 Tic-tac-toe
 Small state space
 Brute force works to play perfectly
 Checkers
 Chinook project built and searched
the complete state space
 Results can be embedded in a
computer player
 Chinook can never be beaten
 Quiz games:
Jeopardy!
 Watson defeated
Jeopardy!
Champions (2011)
 Given a quiz
“answer” and
category:
 Applies multiple kinds
of AI agents to search
database (of
information from the
web)
 Produces 300-500
candidate answers
 Narrows to one answer

REASONING
and evaluates its
certainty in real time
 Scoring and

TASKS THE evaluation are done

in parallel

GAMES WE PLAY
WHAT DO YOU THINK?
 A robot is a physical device
that takes in sensory data
and makes autonomous
responses
Current robot tasks

ROBOTS
 Repetitive or dangerous for
humans

AND
 Manufacturing, bomb disposal,
search-and-rescue

DRONES New research on multiple

cooperating robots
 Schools of robot fish for studying
sea life
 Swarms of robot flies for
reconnaissance
 Groups of robot snowmobiles to
study climate change
 ROBOTS
AND
DRONES
Humanoid robots are
designed for interacting
with people
 Help elderly or hospital
patients
 Monitor small children

Japan is a leader in
humanoid robots
 Aging population needs
support
Asimo, by Honda
 Designed to walk and
move fluidly and robustly
 Can open/close a door to
go through, serve
refreshments, etc.
ASIMO
SOPHIA THE ROBOT
AMICA
Deliberative strategy for robot
control programs
 Maintain detailed internal model of the
world
ROBOT
 Reason about sensory inputs and
choose best response S AND
Reactive strategy DRONE
 Limit/eliminate internal model
 React immediately to sensory inputs S
 Rapid cycle from inputs to responses to
more inputs
WHAT DO YOU THINK?
Drone
 Unmanned Aerial Vehicle (UAV)
 Controlled by a human at a remote site
 Primarily used by military and law
enforcement

Potential uses
 Deliver medical supplies
 Monitor dangerous situations, e.g., fires,
floods, etc.
 Document wildlife
 Document urban traffic

ROBOTS AND
DRONES
SUMMARY
Artificial intelligence programs solve problems in
“intelligent” ways

Knowledge may be represented in many different

ways; choice of representation depends on task

Neural networks simulate the connectionist

structure of the nervous system

Neural networks are trained to produce the correct

responses to inputs

Reasoning may often be state–space search

Swarm intelligence uses colonies of simple agents to solve
problems
Intelligent agents would be artificial personal assistants
Expert systems reason with expert domain knowledge
Game-playing is a common application for AI
Robots perform tedious and dangerous tasks
Drones are unmanned aerial vehicles

SUMMARY