Artificial Intelligence
Introduction
University of California, Berkeley
[These slides are based on those created by Dan Klein and Pieter Abbeel for CS188 Intro to AI at UC Berkeley ([Link]).]
Second Half of Today: What is AI?
▪ What is AI?
▪ What can it do?
▪ What is this course?
AI is having real-world impact
▪ Public imagination
▪ Text assistants
AI is having real-world impact
▪ Public imagination
▪ Text assistants
▪ Image generation
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ $4.8 trillion by 2033
(25x increase in a decade)
[Link]
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
[Link]
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
Bloomberg Law, 2023
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
AI for Scaling Legal Reform: Mapping and Redacting
Racial Covenants in Santa Clara County. Surani et al, 2024.
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
[Link]
AI is having real-world impact
▪ Public imagination
▪ Economy MarketWatch, 2023
▪ Politics
▪ Law
▪ Labor
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
Stanford Digital Economy Lab, 2025
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
The Economist, 2024
▪ Science
▪ Chemistry and biology
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Physics
Wired, 2022
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Conservation
Nature Communications, 2022
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics The New York Times, 2025
▪ Law
▪ Labor
▪ Science
▪ Resource use
Lawrence Berkeley National Lab, 2024
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Health
The Boston Globe, 2018
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Health
▪ Education Forbes, 2023
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Health
▪ Education
▪ Social interaction
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Health
▪ Education
▪ Social interaction The New York Times, 2024
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
▪ Science
▪ Health
▪ Education
▪ Social interaction The New York Times, 2025
AI is having real-world impact
▪ Public imagination
▪ Economy
▪ Politics
▪ Law
▪ Labor
Ok, but what actually is AI???
▪ Science
▪ Health
▪ Education
▪ Social interaction
What should we build?
Should we make machines that…
Think like people? Think rationally?
Act like people? Act rationally?
Rational Decisions
▪ We’ll use the term rational in a very specific, technical way:
▪ Rational: maximally achieving pre-defined goals
▪ Goals are expressed in terms of the utility of outcomes
▪ World is uncertain, so we’ll use expected utility
▪ Being rational means acting to maximize your expected utility
Rational Decisions
▪ We’ll use the term rational in a very specific, technical way:
▪ Rational: maximally achieving pre-defined goals
▪ Goals are expressed in terms of the utility of outcomes
▪ World is uncertain, so we’ll use expected utility
▪ Being rational means acting to maximize your expected utility
A better title for this course might be:
Computational Rationality
Perspectives on Intelligence
▪ Skills-based perspective
▪ “A system is only intelligent if it can do [X].”
▪ Play chess?
▪ Learn from experience?
▪ Use words properly?
▪ Make mistakes?
▪ Not make mistakes?
Perspectives on Intelligence
▪ Embodiment perspective (Rodney Brooks)
Perspectives on Intelligence
▪ Psychometrics perspective (François Chollet)
▪ “Measuring abilities, not skills […] across a broad range of tasks,
including tasks that were previously unknown to the ability-
enabled system and its developers.”
Perspectives on Intelligence
▪ Human-compatible perspective (Stuart Russell)
1. Machine’s objective is to maximize human utility.
2. Initially uncertain about human preferences.
3. Must learn about preferences from human
behavior.
Perspectives on Intelligence
A human being should be able to change a diaper, plan an invasion,
butcher a hog, conn a ship, design a building, write a sonnet, balance
accounts, build a wall, set a bone, comfort the dying, take orders, give
orders, cooperate, act alone, solve equations, analyze a new problem,
pitch manure, program a computer, cook a tasty meal, fight efficiently,
die gallantly. Specialization is for insects.
—Robert A. Heinlein
What About the Brain?
▪ Brains (human minds) are very good at making
rational decisions, but not perfect
▪ Brains aren’t as modular as software, so hard to
reverse engineer!
▪ AI may be better than brains at some tasks
▪ “Brains are to intelligence as wings are to flight”
▪ We can’t yet build AI on the scale of the brain
▪ ~100T synapses in the human brain vs ~1.8T weights
in GPT4
▪ Still, the brain can be a great inspiration for AI!
A (Short) History of AI
▪ 1940-1950: Early days: neural and computer science meet
▪ 1943: McCulloch & Pitts: Perceptron–boolean circuit model of brain
▪ 1950: Turing's “Computing Machinery and Intelligence”
▪ 1950—70: Excitement! Logic-driven
▪ 1950s: Early AI programs, including Samuel's checkers program,
Newell & Simon's Logic Theorist, Gelernter's Geometry Engine
▪ 1956: Dartmouth meeting: “Artificial Intelligence” adopted
“We propose that a 2-month, 10-man study of artificial intelligence be carried out
during the summer of 1956 at Dartmouth College in Hanover, New Hampshire. The
study is to proceed on the basis of the conjecture that every aspect of learning or any
other feature of intelligence can in principle be so precisely described that a machine
can be made to simulate it. An attempt will be made to find how to make machines use
language, form abstractions and concepts, solve kinds of problems now reserved for
humans, and improve themselves. We think that a significant advance can be made in
one or more of these problems if a carefully selected group of scientists work on it
together for a summer.”
A (Short) History of AI
▪ 1940-1950: Early days: neural and computer science meet
▪ 1943: McCulloch & Pitts: Perceptron–boolean circuit model of brain
▪ 1950: Turing's “Computing Machinery and Intelligence”
▪ 1950—70: Excitement! Logic-driven
▪ 1950s: Early AI programs, including Samuel's checkers program,
Newell & Simon's Logic Theorist, Gelernter's Geometry Engine
▪ 1956: Dartmouth meeting: “Artificial Intelligence” adopted
▪ 1969: Minsky & Papert: perceptrons can’t learn XOR/parity!
▪ 1970—90: Knowledge-based approaches
▪ 1969—79: Early development of knowledge-based systems
▪ 1980—88: Expert systems industry booms; backpropagation makes
it feasible to train multi-layer neural networks
▪ 1988—93: Expert systems industry busts: “AI Winter”
▪ 1990—2010: Statistical approaches, agents
▪ Resurgence of probability, focus on uncertainty
▪ Agents and learning systems… “AI Spring”?
▪ 1992: TD-Gammon achieves human-level play at backgammon
▪ 1997: Deep Blue defeats Gary Kasparov at chess
▪ 2002: Embodied AI; Roomba vacuum invented
A (Short) History of AI
▪ 2010—2017: Big Data, GPUs, Deep Learning
▪ 2011: Apple releases Siri
▪ 2012: AlexNet wins ImageNet competition
▪ 2015: DeepMind achieves “human-level” control in Atari games
▪ 2016: DeepMind’s AlphaGo defeats Lee Sedol at Go
▪ 2016: Google Translate migrates to neural networks
▪ 2017—: Scaling Up, Large Language Models
▪ 2017: Google invents Transformer architecture
▪ 2017: DeepStack/Libratus defeat humans at poker
▪ 2018-2020: AlphaFold predicts protein structure from amino acids
▪ 2021-2022: Modern text-to-image generation
▪ 2022: OpenAI releases ChatGPT
▪ 2023: Every other company also releases a chatbot
▪ 2024: Nobel prizes in physics and chemistry go to AI advances
A (Short) History of AI
▪ Zooming out, some general patterns:
▪ Overclaiming followed by disillusionment (AI winters)
▪ Simultaneous discovery, or rediscovery, of ideas
▪ Interdisciplinary
▪ Math, statistics, control theory, neuroscience, psychology, economics,
philosophy, computer engineering, linguistics
▪ And increasingly: law, medicine, education…
What Can AI Do?
Quiz: Which of the following can be done at present?
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem?
Perform a surgical operation?
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
What Can AI Do?
Quiz: Which of the following can be done at present? Google’s AlphaGo beats Lee Sedol at Go in 2016
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem? But … a plot twist in 2023!
Perform a surgical operation?
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
What Can AI Do?
Quiz: Which of the following can be done at present?
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem?
Perform a surgical operation?
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
What Can AI Do?
Quiz: Which of the following can be done at present?
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem?
Perform a surgical operation?
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
What Can AI Do?
Quiz: Which of the following can be done at present?
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem?
Perform a surgical operation?
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
What Can AI Do?
Quiz: Which of the following can be done at present?
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem?
Perform a surgical operation? ChatGPT, 2025
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
What Can AI Do?
Quiz: Which of the following can be done at present?
Win against any human at chess?
Win against the best humans at Go?
Play a decent game of table tennis?
Unload any dishwasher in any home?
Drive safely along the highway?
Drive safely along streets of San Francisco?
Buy a week's worth of groceries on the web?
Buy a week's worth of groceries at Berkeley Bowl?
Discover and prove a new mathematical theorem?
Perform a surgical operation?
Translate spoken Chinese into spoken English in real time?
Win an art competition?
Write an intentionally funny story?
Construct a building?
This Course: Designing Rational Agents
▪ An agent is an entity that perceives and acts.
▪ A rational agent selects actions that maximize
its (expected) utility.
▪ Characteristics of the percepts, environment,
and action space dictate techniques for
selecting rational actions
▪ This course is about:
Sensors
Environment
▪ General AI techniques for a variety of problem Percepts
Agent
types ?
▪ Learning to recognize when and how a new Actuators
Actions
problem can be solved with an existing technique
Pac-Man as an Agent
Agent Environment
Sensors Percepts
?
Actuators Actions
Pac-Man is a registered trademark of Namco-Bandai Games, used here for educational purposes
Course Topics
Core Components of Rational Agents:
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
Course Topics
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
How can I use my model of the world to find a
sequence of actions to achieve my goal?
Course Topics
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
How can I make sense of uncertainty?
Course Topics
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
How can I learn a model of the world from data?
Course Topics
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
How can I learn a policy for any situation
so that I can maximize utility?
Course Topics
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
Course Topics
Search & Probability &
Planning Inference
Supervised Reinforcement
Learning Learning
Applications
Impact on Sciences, Technology, Society
By the end of this course you’ll:
▪ Build and understand math of rational, learning agents
▪ Select and apply the right AI methods for wide range of problems
▪ Recognize how these methods are used in modern AI systems
▪ Be prepared to make decisions on how AI is used in society
Next Lecture: Search