IB Digital Society: 3.6 Artificial Intelligence

AI

Creating a system that produces results comparable to human intelligence.

Artificial Intelligence

Creating a system that produces results comparable to human intelligence.

Artificial Neural Network

AI system that attempts to mimic the neurons and synapses in the human brain.

Boolean logic

Logic in which clauses can have one of two states - such as yes or no, true or false.

Brute force searching

AI technique that considers all possible solutions, looking for the best.

CAPTCHA

System designed to create text that is unreadable to a computer but understandable to a person, to reduce spam.

Inference Chaining

Use of logical statements to come to a conclusion.

Computational intelligence

Approach that tries to create systems that think and learn in the same way humans do.

Confidence interval

Level of certainty in an answer a pattern recognition system provides.

Expert system shell

Software used to create expert systems.

Expert systems

Software designed to make the same decisions that a human expert would, in a given knowledge domain.

Feedback loop

Use of previous answers (right or wrong) to improve the decision making process next time.

Fuzzy logic

Logic in which items can have multiple values. Used in AI.

Fuzzy set theory

System in which items can be partial or complete members of a set. Used in AI.

Hand writing recognition

System to recognise human writing and convert it to text.

Heuristics

General rules for performing a task, used to improve the perform of searching algorithms in AI applications.

IF-THEN rule

Rule used by the inference engine in an expert system to describe the relationship between key concepts.

Inference engine

Part of an expert system which attempts to relate the users input with knowledge stored in the knowledge base.

Inference rule

A rule used by the inference engine in an expert system to describe the relationship between key concepts/items.

Knowledge base

Area of an expert system where all facts about the knowledge domain are stored.

Knowledge domain

Area of knowledge in which an expert system specialises.

Knowledge engineer

Programmer responsible for entering expert knowledge into an expert system.

Logic

Rule used by the inference engine in an expert system to describe the relationship between key concepts.

Machine learning

The extraction of knowledge from data based on algorithms created from training data.

Natural Language Processing (NLP)

Techniques for processing human languages to enable a computer to understand their meaning.

Neural networks

AI technique that tries to simulate the human brain, using neurons and synapses.

Pattern recognition

Computational Intelligence technique where computers are trained on examples and learn to recognise similarities between them.

Searching

AI technique that considers all possible solutions, looking for the best.

Speech recognition

Computer system that can process spoken language and understand its meaning.

Training data

Example data used in a pattern recognition system.

Turing test

Proposed test to see if a computer is intelligent or not.

User interface

Part of an expert system that accepts users inputs and presents answers.

Supervised Learning

Training an AI system using a huge number of examples with specific classifications for each data which is provided by a knowledge engineer.

Unsupervised Learning

When an AI system can look at data on its own and build rules for deciding what it is seeing.

Semi-supervised learning

A type of machine learning where the data used to build models contains data with explicit classifications, but is also free to develop its own additional classifications that may further enhance result accuracy.

Reinforced Learning

Machine learning using trial and error on an unlabeled data set. Learning is gained through positive and negative feedback.

Facial recognition

A biometric technology that looks for unique measurements in an individual's face. Uses AI, specifically semi-supervised learning to perform pattern recognition to authenticate a user.

Regression

A reversion to immature patterns of behavior.

Data mining

The process of analyzing data to extract information not offered by the raw data alone

Disadvantage of Boolean Logic

- Only provides two possible outcomes from given input data - narrowed possibilities.

- Cannot handle uncertainty - not flexible for use outside its design/specialism.

- Relies on complete and precise data to operate.

Advantages of Boolean Logic

- Able to provide quicker results compared to fuzzy logic.

- Higher rates of accuracy, if used within the specialism area.

Advantages of Fuzzy Logic

- Can provide a greater breadth options/outcomes based on probability.

- Able to operate on incomplete and vague data sets.

- Can be used/applied to various context, but with the increased risk of inaccurate results.

Disadvantages of Fuzzy Logic

- Increased risk of inaccurate results due to the incomplete or vague data sets.

- The results of provided the systems may not be extensive, and may be time consuming to review each possibility.

- Increased time to produce results.

Deep Learning

A type of machine learning that uses multiple layers of interconnections among data to identify patterns and improve predicted results. Deep learning most often uses a set of techniques known as neural networks and is popularly applied in tasks like speech recognition, image recognition, and computer vision.

Super AI

Systems that are smarter than the best human brains, and which can make deductions about unknown environments; whether and how these systems will be developed is the subject of intense debate.

Weak/Narrow AI

a machine or system that mimics some aspect of human intelligence; characterized by its relatively narrow range of thought-like ability

Domain-specific AI

Artificial Intelligence that perform tasks better than humans in a certain domain.

Singularity

A hypothetical point in time at which super-intelligent machines become uncontrollable and irreversible, resulting in unforeseeable changes to human civilization.

Cybernetics

The science of communications and automatic control systems in both machines and living things.

Multiplicity

A future in which artificial intelligence and robots are developed to work alongside people, rather than to replace them.