Module 1 - What is AI? Flashcards
What is AI?
What is Artificial Intelligence?
AI is the study of creating machines that think, learn, and act rationally. There are two approaches:
- human-centered AI (mimicking human cognition)
- rational AI (optimal decision-making).
What is rationality in AI?
AI should be rational by making decisions that maximize success.
Human-centered approach: Observing human behavior to model intelligence.
Rationalist approach: Using logic and mathematics to define AI decision-making.
The Turing test & total Turing test
Turing test: A machine is intelligent if a human cannot distinguish it from another human in conversation.
Total turing test: Adds perception (computer vision) and manipulation (robotics) as extra challenges.
What are the six disciplines required to pass the total Turing test?
Natural language processing – understanding text/speech.
Knowledge representation – storing and retrieving facts.
Automated reasoning – drawing conclusions from facts.
Machine learning – improving performance over time.
Computer vision – understanding visual input.
Robotics – moving and manipulating objects.
What is the difference between strong AI and weak AI?
Strong AI: aims to replicate full human intelligence (AGI).
Weak AI: designed for specific tasks (e.g., chatbots, recommendation systems).
In Depth Q. The Turing Test has been criticized as an insufficient measure of intelligence. What are some key objections to it?
Chinese Room argument (Searle) – a machine could pass the test without truly understanding.
Consciousness issue – intelligence might require subjective experiences.
Behavior vs. Thought – passing the test proves imitation, not real thinking.
In Depth Q. Why is achieving perfect rationality in AI often infeasible?
AI systems operate in complex environments where time and computational resources are limited. Instead, AI uses limited rationality by approximating the best decisions under constraints.
What are the main concerns regarding AI safety?
- AI could lead to autonomous weapons.
- AI decisions may become unpredictable and dangerous.
- Experts like Stephen Hawking warn AI could be a threat to humanity.
How does John McCarthy define AI?
The science and engineering of making intelligent machines.”
Why is it difficult to define intelligence in AI?
Intelligence can mean different things (problem-solving, reasoning, adapting, creativity).
Even in psychology, there is no universal definition of intelligence.
Can AI be truly intelligent if it doesn’t have emotions?
Some argue emotions are necessary for decision-making (e.g., humans consider emotions in moral choices).
Others argue rationality alone is enough for intelligence.
What does Stephen Hawking say about AI?
AI could spell the end of the human race
Hawking’s view reflects a significant concern about the potential dangers of AI.
What is Elon Musk’s perspective on AI?
It’s like summoning the demon
Musk emphasizes the unpredictable and potentially harmful consequences of AI.
What does Brett Kennedy believe about robots?
I have first-handed knowledge of how hard it is for us to make a robot that does much of anything
Kennedy’s experience highlights the challenges in AI development.
How does Alan Winfield describe fears of superintelligent robots?
Fears are greatly exaggerated
Winfield suggests that concerns about AI taking over the world may not be justified.
What is David Harding’s view on the debate about AI and machine learning?
Nine parts hype to one part substance
Harding critiques the disproportionate hype surrounding AI compared to its actual achievements.
What is the main idea of Harding’s approach to finance?
Use science to identify and exploit inefficiencies in the markets
Harding advocates for a scientific approach to enhance financial strategies.
What does Theodore Rosak claim about AI’s public deception?
AI’s record of barefaced public deception is unparalleled
Rosak points to a history of exaggerated claims in AI research.
How does John McCarthy define Artificial Intelligence?
The science and engineering of making intelligent machines
McCarthy is one of the pioneers in AI and his definition emphasizes the engineering aspect.
What is Marvin Minsky’s definition of AI?
The science of making machines do things that would require intelligence if done by men
Minsky highlights the goal of AI to mimic human intelligence.
According to John Haugeland, what is AI?
The exciting new effort to make computers think.. machines with minds, in the full and literal sense
Haugeland focuses on the cognitive aspect of AI.
What is Ray Kurzweil’s perspective on AI?
The art of creating machines that perform functions that require intelligence when performed by people
Kurzweil emphasizes the creative aspect of developing intelligent machines.
What is the definition of intelligence according to John McCarthy?
The computational part of the ability to achieve goals in the world
McCarthy’s definition links intelligence to goal achievement.
What does the term ‘artificial’ imply in the context of intelligence?
Non-biological, constructed by humans
The term distinguishes between biological and non-biological forms of intelligence.
What is a key characteristic of the clean robot referred to as Nigel?
Forms a map and finds its way
This behavior indicates that Nigel uses techniques to achieve a specific goal.
Does Nigel learn or improve over time?
No, he doesn’t improve
Nigel operates based on pre-programmed algorithms without learning.
What was the aim of the Dartmouth Summer Research Project on Artificial Intelligence in 1956?
Every aspect of learning or any other feature of intelligence can in principle be so precisely defined that a machine can be made to simulate it
This hypothesis laid the groundwork for future AI research.
Is catching a mouse considered a sort of intelligence?
Yes, it reflects the ability to achieve goals
This suggests that intelligence can manifest in various forms across species.
What does the term ‘hype’ refer to in the context of AI?
Exaggerated claims about AI capabilities
Hype can create unrealistic expectations about what AI can achieve.
According to the text, what is a common skepticism regarding AI?
Too much hype, too less achieved
This reflects a critical view on the actual progress made in AI compared to public perception.
What is AI as a field of research?
AI is part of Cognitive Science.
Why is Cognitive Science considered a collaborative/interdisciplinary field?
Understanding the mind/brain may be too difficult for a single discipline in isolation.
What do researchers in cognitive science seek to understand?
Brain processes as computational systems which manipulate representations.
What does the computational theory of mind propose?
The mind stores symbolic representations and manipulates them via syntactic operations.
How do cognitive scientists view the human mind?
As a complex system that receives, stores, retrieves, transforms, and transmits information.
What is the issue with the fragmentation of AI research?
Researchers don’t always have the same goals or ideas.
How are big AI questions approached in research?
They are often viewed from different perspectives and fields.
What was the trend in AI hype from 1980 to 1996?
AI hype started in 1980 and went down in 1996 due to slow development and high costs.
What happened to machine learning hype during the decline of AI hype?
Machine learning hype went up.
What research area gained attention as AI hype declined?
Neural networks gained attention as more research was done on how they worked.
Fill in the blank: Cognitive scientists view the human mind as a complex system that _______.
receives, stores, retrieves, transforms, and transmits information.
True or False: Most AI research tackles the big AI questions directly.
False.
What is a consequence of the fragmentation of AI research?
Most research is just done on a little part of the big question.
What was the trend for expert systems during the AI hype decline?
Expert systems went down.
What is a primary focus for further AI research based on neural networks?
Modeling after the human brain.
What is Strong AI?
Goal: To understand the human mind/brain as a computational device.
Specific: Focused on the human mind and replicating its functions.
General: Humans can solve a wide range of problems.
Belief: Machines can be built with thought, emotions, and human-like reasoning.
How does Weak AI differto Strong AI in its approach?
It aims to solve problems using AI, without necessarily replicating human-like intelligence.
What is Weak AI?
Goal: To develop intelligent machines that solve problems but are not necessarily human-like.
General: Can involve any AI technique or problem-solving approach.
Specific: Typically focuses on one task (e.g., vision, machine learning).
Doesn’t focus on theories of intelligence—only cares about solving problems.
Why is Strong AI considered “specific” while Weak AI is “general”?
Strong AI is specific because it focuses on human intelligence only.
Weak AI is general because it considers any type of AI technique.
In Depth Q. Could Weak AI eventually lead to Strong AI? Why or why not?
Some argue that advancing weak AI could accumulate knowledge and eventually lead to Strong AI.
Others believe Weak AI lacks consciousness and true understanding, making Strong AI impossible.
In Depth Q. Which AI approach (Strong or Weak) is more useful today? Why?
Weak AI is more useful today because it powers real-world applications like chatbots, self-driving cars, and recommendation systems.
Strong AI is still a theoretical concept, with no known implementations.
What is AI as Science?
Goal: Reverse-engineer the human mind/brain.
The object of interest exists in nature (e.g., studying human intelligence).
AI as a science might fail if it cannot fully understand human intelligence.
Depends on humans to study and model intelligence.
What is AI in Engineering?
Goal: Engineer clever machines, regardless of human cognition.
The object of interest is unknown and must be discovered or created.
AI as engineering has already succeeded in applications like automation and robotics.
Independent of humans—focuses on solving problems, not replicating human thought.
What is an example of AI as science?
Studying how neurons in the brain work to create biologically inspired AI.
What is an example of AI as engineering?
Creating self-driving cars or AI-powered robots without human-like reasoning.
In Depth Q. Can AI as engineering exist without AI as science? Why or why not?
AI as engineering can succeed without science because it focuses on practical solutions (e.g., speech recognition, computer vision).
However, AI science helps improve AI engineering by advancing understanding of cognition and learning processes.
In Depth Q. Why is AI as science considered more uncertain than AI as engineering?
AI as science depends on whether human intelligence can be fully understood—which is still unknown.
AI as engineering already works because machines solve tasks without requiring human-like thought.
