Artificial Intelligence

Artificial intelligence (AI) is the intelligence by machines or software.

AI Application Domains

AI application domains can be considered as the AI goals.

AI application domains featured on this post:

Knowledge Representation and Reasoning (KR&R)
Computer vision
NLP

Knowledge Representation and Reasoning

Knowledge Representation and Reasoning (KR&R) implies both representing and interpreting knowledge.

Computer Vision

Computer vision implies that computer are able to interpret images.

Natural Language Processing (NLP)

Natural Language Processing (NLP) helps computers communicate with humans in their own language and scales other language-related tasks.

Task of NLP include text classification, sentiment analysis, machine translation, named entity recognition, part-of-speech tagging and question answering.

There are different approaches to NLP. The earliest ones (1970s) were rule-based or statistical-based, and are used for parsing, tokenization, or syntactic analysis. The more modern are based on machine learning and LLMs.

NLP components:

Speech processing
Text-to-speech

AI Application Orientations

AI application orientations:

Predictive
Generative
Prescriptive
Descriptive

Predictive AI

Predictive AI is used for forecasting, regression and classification.

Generative AI

Generative AI (GenIA) consist of generating data to produce new content.

It could be considered a subpart of deep learning. It is an alternative to the more traditional Predictive IA.

It became popular to the mass in 2021 after the release of Generative Pre-trained Transformer 3 (GPT-3).

A log probability (logprob) is the probability to use an output using logarithmic functions. Logarithmic functions are used instead of percentages because they are more easily computed.

Recover, Amplify, Generate (RAG) is…

Stable Diffusion is a GenIA algorithm to generate images.

Stable Diffusion at Wikipedia

Prescritive AI

Prescriptive AI recommends optimal actions.

Descriptive AI

Descriptive AI describes and analyzes data.

AI Paradigms

An AI paradigm is an approach to achieve any AI goal.

AI main paradigms:

Symbolic
Subsymbolic
Neuro-symbolic

Symbolic AI

Symbolic AI implies that knowledge is represented as symbols and manipulated by logical rules.

You can read this post about symbolic AI.

Subsymbolic AI

Subsymbolic AI implies that knowledge is encoded in patterns of parameters, not explicit symbols.

Examples are neural networks (including deep learning and LLMs), evolutionary algorithms and statistical models.

Subsymbolic AI techniques:

Machine learning
Reinforcement learning
Probabilistic methods
Evolutionary algorithms

Machine Learning

You can read this post about machine learning.

Evolutionary algorithms

Evolutionary algorithms includes:

Genetic Algorithms (GA)
Genetic Programming (GP)
Evolution Strategies (ES)
Neuroevolution (evolving neural network architectures & weights)

Neural-symbolic AI

Hybrid AI or neural-symbolic AI combines aspects of symbolic and subsymbolic AI.

AI System Architecture

AI system architectures:

Distributed AI
Agentic AI

Distributed AI

A distributed AI refers to an AI architecture where components are distributed along different systems.

Distributed AI goals:

Ensuring agents acts coherently when taking decision or making actions.
Enabling agent reasoning about other agents actions and plans.
Developing platforms for multiagent systems and development methodologies.

Agentic AI

An IA agent is a system that receives information and takes decision autonomously.

Agentic IA is an informal term that implies the use of IA agents that can take autonomous decisions in a system. It is often implemented using reinforcement learning, planning and LLM reasoning.

This term was popularized in the 2020s.