This page provide terminologies and definitions of common technical terms for AI and Machine Learning. This page will be constantly updated as i progress through the topic.
Definition
- Artificial Intelligence is a computer system that simulates the reasoning that humans use to learn from new information and make decisions.
- Based on patterns in existing data and can then learn from its errors to increase its accuracy
Types of AI
- Narrow AI (Weak AI)– Also know as Weak AI. Artificial Intelligence that performs one specific task – though can often outperform a human at that task.
- Artificial General Intelligence (AGI) – Also known as Strong AI or Full AI. In theory, an artificial intelligence that could understand, learn and/or perform any mental or intellectual task a human can do.
As of 2023, AGI is still fictional but companies such as OpenAI (through ChatGPT) and Google (through DeepMind) are working on it.
Machine Learning
One of the subset of AI, it enables machine to learn on its own by analyzing training data.
Common use cases for Machine Learning
- Classification – for identifying patterns and group data
- Regression – for predicting an outcome
Two categories of ML algorithm that should be considered:
- Supervised – An approach wherein training dataset must have both the input data and the desired output
- Unsupervised – Only input data is provided, the machine learn on its own by finding patterns and relationship from the input.
Deep Learning
A subset of machine learning. machine learning techniques that uses neural networks. This tends to be much more expensive the the normal machine learning, but it is more powerful.
Natural Language Processing (NLP)
Area of AI that focuses on recognizing, understanding, analyzing even emulating how humans communicate using either speech or writing text or both.
Sentiment Analysis
A specific, “narrow” application of NLP to analyze and determine the sentiment or emotional tone expressed in text or speech.
Generative AI
AI capable of creating new and original content such as text, images, music, video and computer code.
Examples
- ChatGPT
- Bard
- Bing A.I
These companies uses a different technique called LLM, or Large Language Mode – An AI system extensively trained on vast amounts of text data, to then interact with and generate human language.
Foundation Model
An AI trained on huge amounts of data, to then adapt to multiple uses. Foundation models include text-based LLMs and image-based generative AI.
Hallucinations
Generative AI may create results and answers that look convincing but are entirely invented, with no basis in reality.
