Summary Artificial Intelligence
(Lectures)
Goals:
At the end of the course, the student can:
Describe how the field of artificial intelligence emerged from mathematics, cognitive
psychology, and cybernetics;
Give high-level explanations of several algorithms, including graph traversal, expert system
reasoning, reinforcement learning, and evolutionary algorithms;
Explain how artificial neural networks can be trained and evolved;
Distinguish between the machine learning techniques of supervised, unsupervised, and
reinforcement learning;
Explain how robotics is different from other fields of AI;
Discuss the potential influence of AI on society and work environments.
** means that it was not discussed in the lecture.
1
,Week 1: Introduction and history of artificial
intelligence
Lecture 1 – 28 October 2024:
Main topics of this lecture:
Course organization and practicalities
What is this course about?
History of AI
Weak versus strong AI
Symbolic versus connectionist AI
Course organization and practicalities:
Course objectives over the weeks:
How did AI develop as a science?
What are the different flavours of AI research?
What are some techniques behind intelligent software?
How can robots seem so intelligent?
o There is a distinction between being intelligent and seeming intelligent
How can computer systems learn new things? (machine learning)
How is AI used in practice?
Written exam will consist of 6 essay questions with sub questions. December 19th!!
Exam consists of:
Lectures (slides + extra information given during lecture)
Required readings on brightspace
What is this course about? What is artificial intelligence?:
This course will be about the most important aspects of artificial intelligence. Not just ChatGPT, but
also the history of AI and the developments that led to the current state of the field.
It’s also about the overlap between cognitive psychology and artificial intelligence.
What is AI about?
It is impossible to literally program everything a computer is supposed to say.
How can we program a system that does not only output exactly what you put in there?
How can we program machines, that we program ourselves, that hopefully in the future can do things
that we can’t do?
What is artificial intelligence? If you ask different people, you get different answers.
Distinction between thinking and acting
Distinction between doing something humanly and rationally
o Humanly ≠ rationally.
o Humans are not rational actors, even if the literature pretends that they are.
2
, Thinking Humanly Thinking Rationally
“The exciting new effort to make computers “The study of mental faculties through the use of
think...machines with minds, in the full and computational models.”(Charniak and
literal sense.” (Haugeland, 1985) McDermott, 1985)
“[The automation of] activities that we associate “The study of the computations that make it
with human thinking, activities such as decision possible to perceive, reason, and act.”(Winston,
making, problem solving, 1992)
learning...”(Bellman,1978)
Acting Humanly Acting Rationally
“The art of creating machines that perform “Computational Intelligence is the study of the
functions that require intelligence when design of intelligent agents.” (Poole et al.,1998
performed by people.”
“AI . . . is concerned with intelligent behaviour
“The study of how to make computers do things in artifacts.” (Nilsson, 1998
at which, at the moment, people are better.”
(Rich and Knight, 1991)
Why are (cognitive) psychologists interested in the field of AI?
Cognitive psychology
o The study of the computations that make it possible to perceive, reason and act.
o Whatever is going on inside the brain, we do believe that there are computations
involved.
Artificial intelligence
o The branch of computer science that studies how to build or program computers to
enable them to do what minds can do.
o AI is interested in building systems that resemble or are intelligent.
AI and other scientific disciplines:
AI ≠ psychology
AI ≠ computer science
However, AI draws inspiration from these disciplines:
o AI puts greater emphasis on computation than psychology.
o AI puts greater emphasis on perception, reasoning, and action than computer science.
So why AI in psychology?
AI researchers believe that psychology is one big inverse problem.
o Psychologists observe behaviour and then try to infer the mechanisms that caused that
behaviour. If we only observe the result, there are almost always multiple ways of
realising that result. So there is no way of saying whether or not our theories are
actually true.
o We have a set of observations (behaviour, psychophysiological measurements, EEG,
fMRI, etc.). We then try to infer the processes producing such observations. Such
inferences are limited, and sometimes even impossible to make.
AI can use forward modelling.
o We design a (simple) system, and see how that behaves (animal like behaviour). If we
build these systems ourselves, than we actually do know what causes the behaviour,
because we are the ones who created the systems.
o Examples: cognitive robotics (lecture 4)
3
, o This is where AI and computational psychology meet.
Algorithms:
Thinking algorithmically:
An algorithm is a set of rules that unambiguously defines a sequence of operations.
o An algorithm is a clear and structured sequence of instructions or rules that describe
step by step how a specific problem can be solved. Each step must be precisely
defined, without room for interpretation. This means that an algorithm should yield
the same results for everyone if followed correctly. An algorithm must lead to a
solution or result in a finite number of steps. An algorithm that runs indefinitely is not
usable.
As long as you follow the rules precisely (as long as you follow the algorithm), you could do
it by hand. It will take a long time to do it, but it is not impossible.
Algorithms are often easiest to think about in terms of functions. A function (in computer
science) is a list of instructions that:
o can take input values (arguments)
Arguments can be numbers, lists of numbers, lists of text, etc.
o apply some computations using these arguments (an algorithm)
o return an output value
Output value can be numbers, lists of numbers, lists of text, etc.
o A function takes an input and returns an output.
Problems with algorithms: human language is extremely underdefined. This is why we usually write
down algorithm in pseudocode. It’s natural language like, but it’s very clear and unambiguous.
It’s clear if you know how to read it.
Algorithms:
Finding the number of people in set L (list of people):
num_people ← 0
for each person in L, do
num_people ← num_people + 1
return num_people
Finding the maximum in a set of numbers L:
if (size of list) == 0 return null
largest ← first element in L
for each other element in L, do
if element > largest, then
largest ← element
return largest
How to look at algorithms: Step by step, very unambiguous, describe a sequence of operations that
lead to the correct result.
History of AI:
How did the field of AI develop? Philosophy of mind:
How does the physical brain give rise to the mental mind?
René Descartes (1596–1650): Dualism, because the mind is not physical.
o Dualism: humans have a soul and a body, these are two separate things.
o Cartesian dualism: dualism where the soul is only responsible for the really human
things. Like thinking and reasoning.
4
(Lectures)
Goals:
At the end of the course, the student can:
Describe how the field of artificial intelligence emerged from mathematics, cognitive
psychology, and cybernetics;
Give high-level explanations of several algorithms, including graph traversal, expert system
reasoning, reinforcement learning, and evolutionary algorithms;
Explain how artificial neural networks can be trained and evolved;
Distinguish between the machine learning techniques of supervised, unsupervised, and
reinforcement learning;
Explain how robotics is different from other fields of AI;
Discuss the potential influence of AI on society and work environments.
** means that it was not discussed in the lecture.
1
,Week 1: Introduction and history of artificial
intelligence
Lecture 1 – 28 October 2024:
Main topics of this lecture:
Course organization and practicalities
What is this course about?
History of AI
Weak versus strong AI
Symbolic versus connectionist AI
Course organization and practicalities:
Course objectives over the weeks:
How did AI develop as a science?
What are the different flavours of AI research?
What are some techniques behind intelligent software?
How can robots seem so intelligent?
o There is a distinction between being intelligent and seeming intelligent
How can computer systems learn new things? (machine learning)
How is AI used in practice?
Written exam will consist of 6 essay questions with sub questions. December 19th!!
Exam consists of:
Lectures (slides + extra information given during lecture)
Required readings on brightspace
What is this course about? What is artificial intelligence?:
This course will be about the most important aspects of artificial intelligence. Not just ChatGPT, but
also the history of AI and the developments that led to the current state of the field.
It’s also about the overlap between cognitive psychology and artificial intelligence.
What is AI about?
It is impossible to literally program everything a computer is supposed to say.
How can we program a system that does not only output exactly what you put in there?
How can we program machines, that we program ourselves, that hopefully in the future can do things
that we can’t do?
What is artificial intelligence? If you ask different people, you get different answers.
Distinction between thinking and acting
Distinction between doing something humanly and rationally
o Humanly ≠ rationally.
o Humans are not rational actors, even if the literature pretends that they are.
2
, Thinking Humanly Thinking Rationally
“The exciting new effort to make computers “The study of mental faculties through the use of
think...machines with minds, in the full and computational models.”(Charniak and
literal sense.” (Haugeland, 1985) McDermott, 1985)
“[The automation of] activities that we associate “The study of the computations that make it
with human thinking, activities such as decision possible to perceive, reason, and act.”(Winston,
making, problem solving, 1992)
learning...”(Bellman,1978)
Acting Humanly Acting Rationally
“The art of creating machines that perform “Computational Intelligence is the study of the
functions that require intelligence when design of intelligent agents.” (Poole et al.,1998
performed by people.”
“AI . . . is concerned with intelligent behaviour
“The study of how to make computers do things in artifacts.” (Nilsson, 1998
at which, at the moment, people are better.”
(Rich and Knight, 1991)
Why are (cognitive) psychologists interested in the field of AI?
Cognitive psychology
o The study of the computations that make it possible to perceive, reason and act.
o Whatever is going on inside the brain, we do believe that there are computations
involved.
Artificial intelligence
o The branch of computer science that studies how to build or program computers to
enable them to do what minds can do.
o AI is interested in building systems that resemble or are intelligent.
AI and other scientific disciplines:
AI ≠ psychology
AI ≠ computer science
However, AI draws inspiration from these disciplines:
o AI puts greater emphasis on computation than psychology.
o AI puts greater emphasis on perception, reasoning, and action than computer science.
So why AI in psychology?
AI researchers believe that psychology is one big inverse problem.
o Psychologists observe behaviour and then try to infer the mechanisms that caused that
behaviour. If we only observe the result, there are almost always multiple ways of
realising that result. So there is no way of saying whether or not our theories are
actually true.
o We have a set of observations (behaviour, psychophysiological measurements, EEG,
fMRI, etc.). We then try to infer the processes producing such observations. Such
inferences are limited, and sometimes even impossible to make.
AI can use forward modelling.
o We design a (simple) system, and see how that behaves (animal like behaviour). If we
build these systems ourselves, than we actually do know what causes the behaviour,
because we are the ones who created the systems.
o Examples: cognitive robotics (lecture 4)
3
, o This is where AI and computational psychology meet.
Algorithms:
Thinking algorithmically:
An algorithm is a set of rules that unambiguously defines a sequence of operations.
o An algorithm is a clear and structured sequence of instructions or rules that describe
step by step how a specific problem can be solved. Each step must be precisely
defined, without room for interpretation. This means that an algorithm should yield
the same results for everyone if followed correctly. An algorithm must lead to a
solution or result in a finite number of steps. An algorithm that runs indefinitely is not
usable.
As long as you follow the rules precisely (as long as you follow the algorithm), you could do
it by hand. It will take a long time to do it, but it is not impossible.
Algorithms are often easiest to think about in terms of functions. A function (in computer
science) is a list of instructions that:
o can take input values (arguments)
Arguments can be numbers, lists of numbers, lists of text, etc.
o apply some computations using these arguments (an algorithm)
o return an output value
Output value can be numbers, lists of numbers, lists of text, etc.
o A function takes an input and returns an output.
Problems with algorithms: human language is extremely underdefined. This is why we usually write
down algorithm in pseudocode. It’s natural language like, but it’s very clear and unambiguous.
It’s clear if you know how to read it.
Algorithms:
Finding the number of people in set L (list of people):
num_people ← 0
for each person in L, do
num_people ← num_people + 1
return num_people
Finding the maximum in a set of numbers L:
if (size of list) == 0 return null
largest ← first element in L
for each other element in L, do
if element > largest, then
largest ← element
return largest
How to look at algorithms: Step by step, very unambiguous, describe a sequence of operations that
lead to the correct result.
History of AI:
How did the field of AI develop? Philosophy of mind:
How does the physical brain give rise to the mental mind?
René Descartes (1596–1650): Dualism, because the mind is not physical.
o Dualism: humans have a soul and a body, these are two separate things.
o Cartesian dualism: dualism where the soul is only responsible for the really human
things. Like thinking and reasoning.
4
