Simplicable Guide

33 Types Of Artificial Intelligence

John Spacey, October 31, 2016

Artificial intelligence is technology that is designed to learn and self-improve. It is typically used to solve complex problems that are impossible to tackle with traditional code. In some cases, artificial intelligence research and development programs aim to replicate aspects of human intelligence or alternate types of intelligence that may exceed human abilities in certain respects. The following are common types of artificial intelligence.

Activity Recognition

Determining what humans or other entities such as robots are doing. For example, a car that can see its owner approaching with a heavy bag of groceries may decide to open an appropriate door automatically.

Affective Computing

AI that seeks to read and use emotion.

Artificial Creativity

AI applications in fields that are viewed as creative such as music, art and design. Also applies to human-like abilities such as including humor in a conversation.

Artificial Immune System

Techniques of machine learning that mimic processes of immune system learning and memory.

Artificial Life

Artificial intelligence, science and engineering modeled upon living systems.

Artificial Stupidity

The ability for an artificial intelligence to perform suboptimally in certain contexts. For example, a game bot may adapt its intelligence to the skill of the other players in a game.

Automation

Automation of decisions or physical tasks using machinery such as robots.

Blockhead

A program that simulates intelligence by retrieving information from a data repository. In some cases, products claim to be artificially intelligent as a marketing approach when their software has a tradition design that doesn't learn in a dynamic way.

Bot

A robot that takes no physical form.

Chatterbot

An artificial intelligence that can talk to humans, often over text chat. Typically designed to pass a Turing test.

Committee Machine

A committee machine is a neural network that combines results or learning from other neural networks. It is an approach to scaling AI.

Computer Automated Design

Most computing is currently focused on automation and calculation. Artificial intelligence may support expansion of this role into areas that are thought to require creativity such as design and architecture.

Computer Vision

Analyzing and understanding visual information is a reasonably complex task that typically requires artificial intelligence.

Decision Support System

The use of artificial intelligence to support human decision making. For example, a tool that determines what information you may need in a given situation.

Embodied Agent

An artificial intelligence that has a body either physical or simulated.

Ensemble Learning

A machine learning technique that uses multiple learning algorithms.

Evolutionary Algorithms

A class of algorithms inspired by processes of biological evolution such as reproduction, mutation, recombination and selection.

Friendly Artificial Intelligence

The study of techniques to make AI safe and useful. It is commonly argued that artificial intelligence represents an existential risk.

Intelligence Amplification

AI tools that are designed to augment human intelligence. For example, a tool might identify options for problem solving.

Machine Learning

Algorithms that learn from data are a fundamental difference between AI and traditional software.

Natural Language Processing

The ability to recognize, interpret and synthesize speech.

Neural Network

Artificial neural networks are an AI approach that were originally inspired by biological neural networks. With time their resemblance to biology has decreased and they are now typically based on statistics and signal processing techniques.

Neuro Fuzzy

A neural network designed to think in fuzzy logic.

Recursive Self-Improvement

Software that writes its own code in cycles of self-improvement. Considered somewhat risky.

Self Replicating Systems

Software easily self replicates with operations such as file copy. It is also possible for robots to self-replicate. This has practical applications such as space exploration.

Sentiment Analysis

Tools that determine the general opinion, emotion or attitude in content such as a comment in social media.

Strong Artificial Intelligence

Research into human-like artificial intelligence.

Superhuman

An ability where machines surpass the greatest human achievements. For example, AI can generally beat humans at chess.

Superintelligence

A theoretical AI that surpasses human intelligence in a comprehensive way.

Swarm Intelligence

Large numbers of small self-organizing units that act locally to generate collective intelligence.

Symbolic Artificial Intelligence

An old fashioned approach to AI that restricts computation to a set of symbols that resemble a language of logic.

Technological Singularity

A hypothetical scenario whereby artificial intelligence rapidly improves by many orders of magnitude using techniques such as recursive self-improvement. Generally thought to result in a superintelligence that humans can't understand or control.