Podcast
Questions and Answers
Who is credited as the first to treat economics as a science?
What is the primary focus of neuroscience?
Which psychologist's work is notably linked to the view of the brain as an information-processing device?
What essential components are needed for artificial intelligence to succeed according to the content provided?
Signup and view all the answers
Which historical figure applied the scientific method to the study of human vision?
Signup and view all the answers
What is described as a key characteristic of cognitive psychology?
Signup and view all the answers
Which two scientists are credited with the first work in artificial intelligence?
Signup and view all the answers
Which field explores the relationship between language and thought?
Signup and view all the answers
What is the primary focus of Hebbian learning as demonstrated by Donald Hebb?
Signup and view all the answers
Who is credited with the introduction of the Turing Test and key concepts in artificial intelligence?
Signup and view all the answers
What significant event took place at Dartmouth College in 1956 related to AI?
Signup and view all the answers
How did John McCarthy describe the early years of AI, characterized by numerous successes?
Signup and view all the answers
What was the purpose of the General Problem Solver developed by Newell and Simon?
Signup and view all the answers
What did McCulloch and Pitts propose about networks of neurons?
Signup and view all the answers
Which term was suggested by some as a more precise alternative to 'artificial intelligence'?
Signup and view all the answers
What was a notable characteristic of the primitive computers used during the early years of AI?
Signup and view all the answers
What was a significant characteristic of the DENDRAL program?
Signup and view all the answers
How did MYCIN differ from DENDRAL in terms of rule acquisition?
Signup and view all the answers
What is the primary function of certainty factors in MYCIN?
Signup and view all the answers
What was one outcome of the Heuristic Programming Project (HPP)?
Signup and view all the answers
What was the economic impact of R1, the first successful commercial expert system?
Signup and view all the answers
What significant trend occurred in the AI industry between 1980 and 1988?
Signup and view all the answers
Which methodology was incorporated in later expert systems, influenced by McCarthy’s Advice Taker approach?
Signup and view all the answers
What algorithm was reinvented in the mid-1980s that influences learning problems?
Signup and view all the answers
What was one of John McCarthy's contributions in 1958?
Signup and view all the answers
What did Simon predict would happen within 10 years regarding computers?
Signup and view all the answers
What was a major limitation of early AI systems?
Signup and view all the answers
What term is used to describe general problem-solving methods in early AI research?
Signup and view all the answers
Which of the following inventions helped address the issue of scarce computing resources?
Signup and view all the answers
What is the primary focus of the Advice Taker program?
Signup and view all the answers
What was a common issue faced by early AI systems when solving problems?
Signup and view all the answers
What challenge did researchers face when implementing AI systems for complex problems?
Signup and view all the answers
Which type of agent ignores the percept history and selects actions based solely on the current percept?
Signup and view all the answers
What is a major drawback of table-driven agents?
Signup and view all the answers
Which of the following accurately describes the real-world environment for AI agents?
Signup and view all the answers
In the context of agent functions and programs, what does the term 'architecture' refer to?
Signup and view all the answers
Which agent type builds upon earlier types, adding a model of the world to improve decision-making?
Signup and view all the answers
What advantage does a utility-based agent have over other basic types of agents?
Signup and view all the answers
What function does the INTERPRET-INPUT function serve in a simple reflex agent?
Signup and view all the answers
Which of the following characteristics does NOT apply to a taxi driving environment?
Signup and view all the answers
Study Notes
History of Artificial Intelligence
- Ancient Greeks and others had made contributions to economic thought, but Adam Smith was the first to treat it as a science using the idea that economies can be thought of as consisting of individual agents maximizing their own economic well-being.
- Neuroscience (1861-present) is the study of the nervous system, particularly the brain. One of the great mysteries of science is how the brain enables thought, but it's clear that the brain is somehow involved in thought—for example, strong blows to the head can cause mental incapacitation.
- German physicist Hermann von Helmholtz (1821-1894) and his student Wilhelm Wundt (1832-1920) applied the scientific method to the study of human vision.
- Psychology (1879-present) is the study of how humans and animals think and act.
- The view of the brain as an information-processing device, a principal characteristic of cognitive psychology, can be traced back to the works of William James.
- It is now a common view among psychologists that a cognitive theory should be like a computer program, describing a detailed information-processing mechanism whereby a cognitive function might be implemented.
- Computer engineering (1940-present) has focused on building efficient computers, particularly for artificial intelligence.
- Control theory and Cybernetics (1948-present) explores how artifacts can operate under their own control.
- Linguistics (1957-present) studies the relationship between language and thought.
Early Work in Artificial Intelligence
- The first work on AI was by Warren McCulloch and Walter Pitts (1943), who drew on three sources: knowledge of the basic physiology and function of neurons, formal analysis of propositional logic due to Russell and Whitehead, and Turing's theory of computation.
- McCulloch and Pitts proposed a model of artificial neurons, each characterized as being "on" or "off." A switch to "on" occurs in response to stimulation by a sufficient number of neighboring neurons. The state of a neuron was conceived of as factually equivalent to a proposition which proposed its adequate stimulus.
- McCulloch and Pitts also suggested that suitably defined networks could learn.
- Donald Hebb (1949) demonstrated a simple updating rule for modifying the connection strengths between neurons. His rule, called Hebbian learning, remains an influential model.
- Hebb demonstrated that any computable function could be computed by some network of connected neurons, and that all the logical connectives could be implemented by simple net structures.
Alan Turing and the Turing Test
- Alan Turing's vision has been the most influential in AI. He lectured on the topic as early as 1947 and articulated a persuasive agenda in his 1950 article "Computing Machinery and Intelligence."
- Turing introduced the Turing Test, machine learning, genetic algorithms, and reinforcement learning.
The Birth of Artificial Intelligence
- John McCarthy, an influential figure in AI, helped establish the field at Dartmouth College in the summer of 1956. He organized a workshop that brought together U.S. researchers interested in automata theory, neural nets, and the study of intelligence.
- The Dartmouth workshop in 1956 was the first official usage of McCarthy's term "artificial intelligence."
Early Enthusiasm and Expectations
- The early years of AI were full of successes in a limited way. Given the primitive technology of the time, it was astonishing whenever a computer did anything remotely clever.
- Researchers responded to the prevailing belief that "a machine can never do X" by demonstrating that computers could do X. McCarthy referred to this period as the "Look, Ma, no hands!" era.
General Problem Solver
- Newell and Simon’s early success was followed up with the General Problem Solver (GPS).
- GPS was designed to imitate human problem-solving protocols and, within a limited class of puzzles, it turned out that the order in which the program considered subgoals and possible actions was similar to that in which humans approached the same problems.
- GPS was probably the first program to embody the "thinking humanly" approach.
LISP
- John McCarthy defined the high-level language LISP in 1958, which became the dominant AI programming language for 30 years.
- That same year, McCarthy invented time-sharing in response to the limited access to scarce and expensive computing resources.
- McCarthy also published “Programs with Common Sense” - the Advice Taker, a hypothetical program that can be seen as the first complete AI system. It was designed to use knowledge to search for solutions to problems.
A Dose of Reality
- AI research was initially overconfident in the potential of early AI systems to solve large, complex problems. This overconfidence stemmed from the promising performance of early AI systems on simple examples.
- Most early programs had limited knowledge of their subject matter and succeeded by means of simple syntactic manipulations.
- Many of the problems AI was attempting to solve proved intractable.
- There were also fundamental limitations in the basic structures being used to generate intelligent behavior.
Knowledge-Based Systems: The Key to Power?
- The early picture of problem solving in AI research was of a general-purpose search mechanism trying to string together elementary reasoning steps to find complete solutions.
- These so-called "weak methods" could not scale up to large or difficult problem instances.
- The alternative was to use more powerful, domain-specific knowledge that allows larger reasoning steps and can more easily handle typically occurring cases in narrow areas of expertise.
DENDRAL
- The DENDRAL program (Buchanan et al., 1969) was an early successful knowledge-intensive system that used a large number of special-purpose rules for its expertise.
- This system demonstrated the clean separation of the knowledge (in the form of rules) from the reasoning component, a key element of McCarthy's Advice Taker approach.
Expert Systems
- Following the success of DENDRAL, the Heuristic Programming Project (HPP) investigated the extent to which the new methodology of expert systems could be applied to other areas of expertise.
- The next major effort in expert systems was in the area of medical diagnosis.
- Feigenbaum, Buchanan, and Dr. Edward Shortliffe developed MYCIN to diagnose blood infections.
- With about 450 rules, MYCIN was able to perform as well as some experts and considerably better than junior doctors.
- MYCIN differed from DENDRAL in two key aspects:
- The rules in MYCIN were acquired from extensive interviewing of experts, not from a general theoretical model.
- The rules had to reflect the uncertainty associated with medical knowledge.
- MYCIN incorporated a calculus of uncertainty called certainty factors (at the time believed to fit well with how doctors assessed the impact of evidence on diagnosis).
AI Becomes an Industry
- The first successful commercial expert system, R1, began operation at the Digital Equipment Corporation in 1982.
- This program helped configure orders for new computer systems and by 1986 was saving the company an estimated $40 million a year.
- The AI industry boomed from a few million dollars in 1980 to billions of dollars in 1988, including hundreds of companies building expert systems, vision systems, robots, and specialized software and hardware.
The Return of Neural Networks
- In the mid-1980s, at least four different groups reinvented the back-propagation algorithm first found in 1969 by Bryson and Ho.
- This algorithm has become a standard tool for training neural networks and has found wide application in machine learning problems in computer science and psychology.
Environment Types and Agent Design
- The environment type largely determines the agent design.
- The real world is partially observable, stochastic, sequential, dynamic, continuous, and multi-agent.
Agent Types
- There are four basic types of agents in order of increasing generality:
- Simple reflex agents
- Model-based reflex agents
- Goal-based agents
- Utility-based agents
Simple Reflex Agents
- The simplest kind of agent is the simple reflex agent.
- These agents select actions based on the current percept, ignoring the rest of the percept history.
- For example, a vacuum cleaner agent that has a rule "If the current location is dirty, then suck."
- A simple rule-based system with a limited number of rules and a simple response to a situation.
Table-Driven Agent
- A table-driven agent maps every possible percept sequence to an action.
- Drawbacks:
- Building and storing a huge table is difficult and time-consuming.
- Lack of autonomy.
- Learning the table entries takes a long time.
Vacuum Cleaner Agent
- A basic vacuum cleaner agent can be programmed with specific rules and a limited set of actions.
- For example, if the current location is dirty, then suck.
- If the current location is clean and the previous location was dirty, then move to the previous location.
The Job of AI
- The job of AI is to design an agent program that implements the agent function, which maps percepts to actions.
- We assume this program will run on some sort of computing device with physical sensors and actuators.
- This computing device is called the architecture.
- An agent is the combination of architecture and program.
- One agent function (or a small equivalence class) is rational. The aim is to find a way to implement the rational agent function concisely.
Studying That Suits You
Use AI to generate personalized quizzes and flashcards to suit your learning preferences.
Related Documents
Description
Explore the development of artificial intelligence through historical perspectives, from ancient contributions to modern neuroscience and psychology. This quiz delves into key figures and concepts that have shaped our understanding of intelligence and thought processes. Test your knowledge on how these diverse fields intersect in the evolution of AI.
