Your browser doesn't support the features required by impress.js, so you are presented with a simplified version of this presentation.
For the best experience please use the latest Chrome, Safari or Firefox browser.
CIS 115
Lecture 17: Artificial Intelligence
Dr. William Hsu
What is Artificial Intelligence?
John McCarthy (1927 - 2011)
Image Source: Wikipedia
Dartmouth Summer Research Project on Artificial Intelligence - 1956
Image Source: Dartmouth
What is Artificial Intelligence?
"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."
- John McCarthy, 2007
Source: Stanford
Yes, but what is intelligence?
"Intelligence is the computational part of the ability to achieve goals in the world. Varying kinds and degrees of intelligence occur in people, many animals and some machines." - John McCarthy, 2007
Source: Stanford
Isn't there a solid definition of intelligence that doesn't depend on relating it to human intelligence?
"Not yet. The problem is that we cannot yet characterize in general what kinds of computational procedures we want to call intelligent. We understand some of the mechanisms of intelligence and not others." - John McCarthy, 2007
Source: Stanford
Intelligent or Merely Human?
Image Source: Wikipedia
Strong AI vs. Weak AI
- Strong AI - matches human intelligence and is capable of performing any task
- Weak AI - only designed to perform a specific subset of intelligent actions
Intelligent Agents
- Agent: Definition
- Any entity that perceives its environment through sensors and acts upon that environment through effectors
- Examples (class discussion): human, robotic, software agents
Agents
- Perception: Signal from environment, may exceed sensory capacity
- Sensors: Acquires percepts, possible limitations
- Action: Attempts to affect environment, usually exceeds effector capacity
- Effectors: Transmits actions, possible limitations
Generic Intelligent Agent Model
Image Source: Dr. Hsu
How Agents Should Act
A rational agent:
- Does the right thing
- Given what it believes
- From what is perceives
Maximum Success Measure (Utility)
- What is the right thing
- How and When to evaluate success
How Agents Should Act
Questions raised about Beliefs, Uncertainty, Knowledge
- Representation
- Reasoning
- Learning
Studying Rational Behavior Helps Understand Decision-Making
Logic Theorist
Image Source: Wikipedia
Allen Newell Herbert Simon
Rational Agents
"Doing the Right Thing"
- Committing actions: limited effectors, in context of agent knowlege
- Meeting Specification: pre/post conditions
Rational Agents Capabilities
- Choice: select actions and carry them out
- Knowledge: represent knowledge about environment
- Perception: capability to sense environment
- Criterion: performance measure to define degree of success
Rational Agents Capabilities
Possible additional capabilities
- Memory: internal model of the state of the world
- Knowledge about effectors, reasoning process, reflexive reasoning
What is Rational?
Criteria
- Determines what is rational at any given time
- Varies with agent, environment, situation
Performance Measure
- Specified by outside observer or evaluator
- Applied (consistently) to (one or more) IAs in a given environment
What is Rational?
Percept Knowledge
- Definition: entire history of percepts gathered by agent
- NB: agent may or may not have state (memory)
Agent Knowledge
- Of environment - "required"
- Of self (reflexive reasoning)
What is Rational?
Feasible Action
- What can be performed?
- What does the agent believe it can attempt?
Structure of Intelligent Agents
Agent Behavior
- Given: a sequence of percepts
- Return: IA's actions
- Simulator: description of results of actions
- Real-world system: committed action
Structure of Intelligent Agents
Agent Programs
- Functions that implement the program
- Assumed to run in a computing environment (architecture)
- Agent = architecture + program
Structure of Intelligent Agents
Applications
- Decision Making: data mining, analytics ("big data"), informatics
- Natural Language Processing (NLP): conversation, speech, text
- Vision
- Robotics and agents
Example: Game Playing Agent
Image Source: Dr. Hsu
Methods for Developing AI
- Knowledge Representation
- Search
- Expert Systems & Knowledge Bases
- Planning: classical, universal
- Probabilistic reasoning
- Machine learning: neural networks, evolutionary computing
- Applied AI: agents focus
- Special topics (NLP focus)
PEAS Framework
Performance Measure
- Specified by outside observer or evaluator
- Applied (consistently) to (one or more) IAs in given environment
Environment
- Reachable Status
- "Things that can happen"
- "Where the agent can go"
- To be distinguished from: observable states
PEAS Framework
Actuators
- What can be performed
- Limited by physical factors and self-knowledge
Sensors
- What can be observed
- Subject to error: measurement, sampling, postprocessing
1 - Simple Reflex Agents
Image Source: Dr. Hsu
2 - Reflex Agents with Memory (State)
Image Source: Dr. Hsu
3 - Goal-Based Agents
Image Source: Dr. Hsu
4 - Utility-Based Agents
Image Source: Dr. Hsu
Artificial Neural Networks
Image Source: Wikipedia
Neural Network Activity
- Each person has 10 images
- Vote if that image is a CAT or DOG
- If you are correct, you get more votes next time
- If you are incorrect, you get less votes next time
Camouflaging Tanks
- 100 photos of tanks behind trees
- 100 photos of just trees
- It worked for all pictures that were used to train the system
- It didn't work for another set of pictures
Why?
Artificial Intelligence Today
Deep Blue
Beat Gary Kasparov at Chess in 1997
Image Source: Wikipedia
Other Uses
- Microsoft Kinect
- Apple Siri
- Google
- Wolfram Alpha
Almost Everywhere!
What we talked about
- Alan Turing & the Turing Test
- John Searle & the Chinese Room
- Newell & Simon's Logic Theorist
- Dartmouth Research Project
- Subtopics & Tools in AI
- Marvin Minsky & Neural Networks
- AI Today (breifly)
What we didn't talked about
- Philosophical Implications
- Ethical Implications
- Solvability - Is there something a human can do that can't be done by an AI?
- The Singularity
- What is "consciousness"?
Assignments
- Read and be prepared to discuss:
- Tubes Chapter 7: Where Data Sleeps
- Blog 8: Artificial Intelligence Everywhere - Due 3/30 10:00 PM
- HTML & CSS Project - Due 3/31 10:00 PM
Blog 8: Artificial Intelligence Everywhere
In today's world, it seems like artificial intelligence is everywhere. From the games we play, to the websites we use, to the systems governing our traffic and energy supply, artificial intelligence is everywhere. For this blog post, choose one example of artificial intelligence you interact with on a regular basis, and tell us about it. Some things you can include in your article:
- Who created it (or the first example of such a system)?
- How does it work? What algorithms or techniques are being used?
- What makes this system useful? Does it have any negative factors?
- How would things be different without this system?
- Is this system useful? Necessary? Overkill? Dangerous? Frivolous?
