Center for Analysis and Design of Intelligent Agents

T-720-ATAI-2019

Lecture Notes, W1

What this Course Is / Is not

Intelligence	This course is about phenomenon we refer to as “intelligence”. A number of features of natural intelligence remain unexplained. In this course it is the question “How can we create a machine that is (at least) as smart as a human?”
AGI	The word “artificial” means “created (from the ground up) by us”. “General” means not specifically targeted to any particular use, but rather to any and every use, as much as that is possible. Many terms have been used to refer to the various aspects that people study wrt intelligence. We use the term “artificial general intelligence” or “general machine intelligence” in the most general sense of these terms (no pun intended), to refer to the various aspects of intelligence that allows it to deal with variety, incompleteness, and incremental information gathering.
Advanced topics	The main focus of course is not the latest and greatest methods to come out of the field called “AI”. However, we will make some references to such methods along the way, and you may even learn something about them. But that is not what is meant by “advanced”.
What does “advanced” mean here?	It refers to advancement toward a deeper understanding of the phenomenon of intelligence, and specifically, cutting-edge and bleeding-edge ideas about how to achieve AGI. Some of the topics we cover are “fringe” in that not very many researchers are studying them. That is, however, of course not any ground on which to dismiss them. If that were the case we would not have discovered DNA or the principles of evolution in nature. Many of the topics are at the border between research fields, some even squarely in the domain of philosophy. The majority, however, are cast in terms of concepts from computer and information science.
History	The phenomenon of intelligence has been studied for ages. Some of the early notable contributions were the Greek philosophers' musings on reasoning and logic. This is not a history course, but we must make some references to the history of philosophy, AI, cybernetics and computer science along the way.

Course Overview

Lectures	Every week involves lectures. Listening is good. Listening and asking questions is better. Reading the assignments, listening, and asking questions is best.
Software Assignments	To enable you to get insight into principles of machine learning and the advanced topics we cover.
Short Essay	To allow you to delve into a topic of particular interest to you.
Pair Software Project	To give you a bit more in-depth experience in programming AGI systems.
Final Exam	To try to gauge how much of the material you actually understand – how much of it you have “ingested”.

Lines of Research & High-Level Topics

Intelligence	Phenomenon. Intelligence is a natural phenomenon, but may have more forms than the examples from nature (the only one that everyone agrees on to call 'intelligent': humans).
Natural Intelligence	Phenomenon. Some kinds of animals are considered “intelligent”, or at least some behavior of some individuals of an animal species other than humans are deemed indicators of intelligence.
Cognitive Science	The study of natural intelligence, in particular human.
Artificial Intelligence	The study of how to make intelligent machines.
Intelligent machines	Systems created by us to display (some or all) features deemed 'intelligent' when encountered in nature.
How to define 'intelligence'	Many definitions have been proposed, see e.g.: A Collection of Definitions of Intelligence by Legg & Hutter.
Definitions: a word of caution	We must be careful when it comes to definitions – for any complex system there is a world of difference between decent definitions and good accurate appropriate definitions.
Related quote	Aaron Sloman says: “Some readers may hope for definitions of terms like information processing, mental process, consciousness, emotion, love. However, each of these denotes a large and ill-defined collection of capabilities or features. There is no definite collection of necessary or sufficient conditions (nor any disjunction of conjunctions) that can be used to define such terms.” (From Architectural Requirements for Human-like Agents Both Natural and Artificial by A. Sloman)

Historical Concepts

AI	In 1956 there was a workshop at Dartmouth College in the US where many of the field's founding fathers agreed on the term to user for their field, and outlined various topics to be studied within the field.
GOFAI	“Good old-fashioned AI” is a term used nowadays to describe the first 20-30 years of research in the field.
Cybernetics	Going back to WWI the field of cybernetics claimed a scope that could easily be said to subsume AI. Many of the ideas associated with information technology came out of this melting pot, including ideas by von Neumann. However, cybernetics has since all but disappeared. Why?
AGI	“Artificial general intelligence”: What we call the machine that we hope to build that could potentially surpass human intelligence at some point in the future – a more holistic take on the phenomenon of intelligence than the present mainstream AI research would indicate. Will we succeed? Only time will tell.

Key Concepts in AI

Problem	A state of affairs, condition, or set of constraints that are desired to be changed.
Goal	The resulting state after a successful change.
Task	A problem that is assigned to be solved by an agent.
Environment	The constraints that may interfere with achieving a goal.
Plan	The partial set of actions that an agent assumes will achieve the goal.
Planning	The act of generating a plan.
Knowledge	Information that can be used for various purposes.
Agent	A system that can sense and act in an environment to do tasks. https://en.wikipedia.org/wiki/Intelligent_agent

Important Concepts in This Course

Methodology	Present methods in AI will not suffice for addressing the full scope of the phenomenon of intelligence, as see in nature.
Attention	The ability to manage resources, including computational (thought), information from the external environment, energy, and time.
Meta-Cognition	The ability of a system to reason about itself.
Reasoning	The application of logical rules to knowledge.
Learning	Acquisition of knowledge that enables more successful completion of tasks.
Life-long learning	Incremental acquisition of knowledge throughout a (non-trivially long) lifetime.
Transfer learning	The ability to transfer what has been learned in one task to another.
Autonomy	The ability to do tasks without interference / help from others.
Constructionist AI	Methodology that relies heavily on human coding for building intelligent systems.
Constructivist AI	Methodology that relies on systems acquiring their own knowledge.

AI is a Broad Field

AI spans many fields	Psychology, mathematics and computation, neurology, philosophy.
Is AI a subfield of computer science?	Yes and No. Yes, because this is the field that has the best and most tools for studying it as a phenomenon. No, because the field does not address important concepts and features of intelligence.
Terminology	Many key terms in AI tend to be overloaded. Others are very unclear. Examples of the latter include: intelligence, agent, concept, thought. Many terms have multiple meanings, e.g. reasoning, learning, complexity, generality, task, solution, proof. Yet others are both unclear and polysemous, e.g. consciousness. One tendency for creating multiple meanings for terms is the habit of, in the beginning of a new research field to use terms to refer to general concepts in nature, and then as time passes to use the same terms to refer to work already done in the field (e.g. expert system, machine learning, neural net). Needless to say, this regularly makes for some lively but useless debates on many subjects.

Some Interesting Case Stories of Intelligence

The Crow	One crow was observed, on multiple occasions, to make its own tools.
The Parrot	One parrot was taught multiple concepts in numbers and logic, including the meaning of “not” and the use of multi-dimensional features to dynamically create object groups for the purpose of communicating about multiple objects in single references. Oh, and the parrot talks.
The Ape	One ape was taught to use sign language to communicate with its caretakers. It was observed creating compound words that it had never heard, for purposes of clarifying references in space and time.

Some (Pressing?) Questions

Isn't AI almost solved?	Short answer: No! If it's almost solved it's been “almost solved” for over 60 years.
Is the Singularity near?	Short answer: Who's to say? Predictions are difficult, especially wrt the future.
Should we fear AI?	AI is a technology. Like any other knowledge it can be used for good or bad. The threat lies with humans, not with machines – human abuse of knowledge goes back to the stone age. By the time the course is finished you will be in a good position to make up your own mind about this.
Is AGI Possible?	Short answer: Yes, in principle. The Devil is in the details.
How long until we have AGI?	Short answer: Nobody knows. Again, by the end of this course you'll be in a good position to answer this on an informed basis.

EOF

Table of Contents

T-720-ATAI-2019

Lecture Notes, W1

What this Course Is / Is not

Course Overview

Lines of Research & High-Level Topics

Historical Concepts

Key Concepts in AI

Important Concepts in This Course

AI is a Broad Field

Some Interesting Case Stories of Intelligence

Some (Pressing?) Questions