Artificial Intelligence

AI Fundamentals

Artificial Intelligence (AI) refers to the capacity of computer systems to execute tasks that traditionally necessitate human intelligence, including learning, problem-solving, decision-making, and understanding language. Rather than following explicit step-by-step programming for every possible scenario, AI systems are "trained" using vast quantities of data to recognise patterns and generate predictions.

The field is highly interdisciplinary, drawing from computer science, data analytics, statistics, hardware and software engineering, linguistics, neuroscience, philosophy, and psychology,. Fundamentally, all AI techniques depend on three core pillars: data, algorithms, and computational power.

The process by which an AI learns typically follows a three-step cycle:

Training Data: The system is fed millions of examples, such as text or images.
Pattern Matching: The AI identifies relationships within that data to construct a "model".
Prediction: When presented with new, unseen data, the AI uses its trained model to make an informed output (e.g., identifying an image as a "dog").

Types of AI

AI is categorised based on its capabilities, ranging from existing technology to theoretical future developments:

Artificial Narrow Intelligence (ANI): This is the only form of AI that currently exists. These systems are designed for specific tasks, such as playing chess, driving a vehicle, or engaging in chat. ANI does not possess self-awareness or reasoning; it functions within predefined parameters using data and algorithms. However, it carries risks, such as biased outputs if the training data is poor.
Artificial General Intelligence (AGI): A theoretical future step where AI would possess human-level intelligence across a broad range of tasks. AGI would be autonomous, adaptive, and capable of using human-like reasoning to learn and improve.
Artificial Superintelligence (ASI): The most advanced theoretical form of AI, representing a self-aware entity that operates beyond human control. It would significantly surpass human intelligence in areas like creativity, reasoning, and emotional intelligence. Some researchers suggest ASI could pose an existential threat to humanity.

Artificial Intelligence vs Machine Learning vs Deep Learning

Artificial Intelligence (AI): AI is the broadest field, encompassing any technology that mimics human cognitive functions. It encompasses various disciplines, including Natural Language Processing (NLP) and Computer Vision.
Machine Learning (ML): A subset of AI, Machine Learning focuses on training computers to learn from data to make decisions without being explicitly programmed for specific tasks.
- Learning Styles: It can be supervised (it is the simplest, using labelled data), unsupervised (finding patterns in unlabeled data), semi-supervised (a hybrid approach), or reinforcement learning (learning through rewards and penalties).
- Techniques: Common algorithms include linear regression, decision trees, random forest, support vector machines (SVMs), and k-nearest neighbours (KNN).
Deep Learning (DL): A more advanced subset of ML, Deep Learning utilises "neural networks" inspired by the human brain. While classic ML models may have only one or two hidden layers, deep neural networks contain at least three, and often hundreds, of layers. This allows the system to process data in multiple layers and automate the extraction of complex patterns from massive, unstructured datasets without human intervention.
Generative AI (Gen AI): Is a specific type of Deep Learning that creates entirely new content (such as text, images, or code) rather than just analysing or categorising existing data.
- Foundation Models: Gen AI starts with a "foundation model", a deep learning model trained on petabytes of data from the internet.
- Large Language Models (LLMs): These are the engines behind text-based Gen AI (like ChatGPT), trained to understand and manipulate human language.
- Mechanism: These models encode representations of patterns and relationships in data and then draw from these representations to generate original work in response to user prompts.

Page updated

Google Sites

Report abuse