Definitions

What Is AI - AI is a simulation of human intelligence processes by computer systems.

John McCarthy - Father of Artificial Intelligence.
- “The science and engineering of making intelligent machines”
Alan Turing - Father of modern computer science.
- Introduced in his seminal 1950 paper, Computing Machinery and Intelligence, originally
framed as the "Imitation Game"
The Turing Test - A machine exhibits intelligent behavior equivalent to, or indistinguishable from, that of a
(Artificial general human.
intelligence)
Innovative AI - Applications of artificial intelligence that transform workflows, create autonomous
systems, and redefine human creativity
AI reasoning - AI reasoning is the process of using available information, rules, or conditions to reach a
conclusion or decide what action to take.
Environment - The environment is the world or situation in which the agent operates.
Sensor - Sensors receive data from the environment.
Actuator - Actuators allow the agent to act on the environment.
Perceive - The agent collects information from the environment using its sensors.
Think - The agent processes the information and decides what to do.
Act - The agent executes a command using its actuators.
Agent An agent is anything that perceives its environment and acts upon it.
Rational Agent - A rational agent is an agent that selects actions that maximize the probability of achieving
its goal.
Simple Reflex Agent -A Simple Reflex Agent responds directly to the current situation.
Model-Based Reflex -A Model-Based Reflex Agent keeps an internal model/state of the environment.
Agent
Goal-Based Agent -A Goal-Based Agent considers a goal and plans actions toward achieving it.
Utility-Based Agent - A utility-based agent evaluates all possible ways to achieve goal outcomes and chooses
the one with the highest utility according to its preferences.
Learning Agent - A Learning Agent can learn from experience and improve its behavior over time.
Bots - A bot is an agent that exists purely in software.
Rational Bot - A bot is a Rational Agent when its responses maximize the probability of achieving the
user's goal.
Stochastic - A stochastic environment is one where the same action does not always produce the
Environment same result.
Linguistic Agent - A linguistic agent is an agent that perceives and processes human language, identifies the
(e.g., bots) user's intent, and produces an appropriate response or action.
Chinese Room - The setup: Imagine you are locked in a room. You do not speak or understand Chinese.

(A famous 1980 - The rules: You have a large English rulebook that tells you how to match incoming Chinese
thought experiment characters with outgoing Chinese characters.
by philosopher John
Searle) - The interaction: People outside slide Chinese symbols under the door. You follow the
rulebook to select correct matching symbols and slide them back.

- The illusion: To the people outside, the room responds so well that it seems fluent in
Chinese.

- The reality: You still do not understand a single word of Chinese; you are just blindly
following mechanical instructions

Weak AI - refers to AI that acts as if it is intelligent but does not necessarily understand the meaning.
Strong AI - refers to the idea that a machine can actually possess a mind and genuine understanding.
Intent mapping - The process of mapping a user's utterance to the underlying intent or goal that the user
wants to accomplish.
PEAS Framework - PEAS is a framework for specifying an intelligent agent by defining its Performance
measure, Environment, Actuators, and Sensors.
Performance - Defines how the success of the AI agent is measured.
Measure
Environment - Defines the environment in which the AI agent operates.
Actuators - Defines what the AI agent can control or change.
Sensors - Defines what information the AI agent can receive.
AI Problem - In AI, a "problem" is a situation where an agent must choose actions to reach a specific
goal.
Initial State - Where the agent starts.
Actions - Possible actions the agent can take.
Goal Test - Checks whether the desired goal has been reached.
Search in AI - Search is the process of exploring the "State Space" to find a path from an initial state to a
goal state.
Transition Model - describes what happens when an action is performed.
Path Cost - measures the cost of taking a particular sequence of actions.
State Space - State space is the collection of all possible states that an AI agent can reach while solving
a problem.
Search Tree - A branching representation of possible paths from a starting point.
Search Graph - A network representation of states and their relationships, which can include shared
states and cycles.
Breadth-First - Explore all immediate neighbors first, then move to the next "depth."
Search (BFS)
Depth-First Search - A search strategy that explores one branch deeply before moving to other branches.
(DFS)
Symbolic AI - Symbolic AI is an AI approach that represents knowledge explicitly using symbols, rules,
and logic written by humans.
Machine Learning - Machine Learning is a branch of AI where computers learn patterns or rules from data and
examples to improve performance or make predictions.
Supervised - The model learns from labeled data
learning
Unsupervised - The model finds patterns in unlabeled data.
learning
Reinforcement - The model learns by trial and error, getting rewards or penalties
learning
Problem Definition - Identifying the objective (classifying an email as spam or non-spam).
Data Collection - Gathering a representative dataset
Preprocessing - This involves "cleaning" the data (handling missing values, removing noise, and
normalizing data scales)
Algorithm Selection - Choosing a model that best suits the data structure.
Training - The process where the algorithm "sees" the data and adjusts its internal parameters to
minimize error.
Evaluation - Testing the model on data it has never seen to ensure it can generalize to new situations
rather than just memorizing the training set.
Model Deployment: Implementing the final "rulebook" into a real-world application.
Knowledge - The process of representing information in a form that an AI system can understand and
Representation use.
Feature Engineering - The process of selecting or creating useful measurable features from data for a machine-
learning model.
Clustering - An unsupervised machine learning technique that groups unlabeled data points into
clusters based on shared characteristics or similarity measures.
Bias & Fairness - Biased training data can cause the ML model to make unfair decisions.
Privacy Risks - Large datasets may contain sensitive information, so privacy protection is important.
Lack of - Complex ML models can be difficult to understand or interpret, reducing trust and
Explainability accountability.
Data Quality - The quality of the training data strongly affects the quality of the AI system.
Probability in - Probability is used in language because human language is ambiguous.
language
AI uses probabilistic models to determine the most likely meaning or continuation based
on context, dealing with degrees of belief rather than absolute facts.
Natural Language - NLP is a field at the intersection of AI and linguistics that enables computers to
Processing (NLP) understand, interpret, and generate human language.
Natural Language - Converts human language into a machine-readable form and extracts things such as
Understanding keywords, emotions, relations, and semantics.
Natural Language - Converts computerized data into a natural-language representation.
Generation
Tokenization - Breaking text into smaller units called tokens, such as words, phrases, or symbols.
Sentiment Analysis - Identifying the emotional or subjective state in text and classifying it as Positive, Negative,
or Neutral.
Named Entity - Identifying and classifying entities such as people, organizations, locations, and time
Recognition (NER) expressions.
Machine Translation - Converting text from one language to another using probabilistic alignment and examples
from a corpus.
Chatbots - Using language processing and probability to select an appropriate response to the user's
input.
Basic NLP Tasks - Basic NLP tasks/applications include Tokenization, Sentiment Analysis, Named Entity
/Applications Recognition (NER), Machine Translation, and Chatbots.
Socio-ethical - Socio-ethical considerations in AI are the ethical and societal issues caused by the
considerations in AI development and use of AI, including bias and fairness, employment, understanding,
transparency, inclusiveness, and accountability.
Fairness - Fairness in AI means ensuring that AI systems treat people and groups fairly and do not
produce unjust or discriminatory outcomes.
Inclusiveness - Inclusiveness in AI means ensuring that AI systems are designed to be accessible and
beneficial to diverse groups of people without unfairly excluding anyone.
Transparency - Transparency in AI means making the AI system's processes, decisions, and outcomes
understandable and clear to users and stakeholders.
Accountability - Accountability in AI means ensuring that people or organizations are responsible for the
decisions, actions, and impacts of AI systems.
Prompt Engineering - Prompt Engineering is the process of designing effective prompts or instructions to guide
an AI system toward producing the desired output.
Role Prompting - Role prompting is a prompt-engineering technique where a specific role is assigned to the
AI to guide its behavior and response.
Few-Shot Prompting - Few-shot prompting is a technique where a small number of examples are provided in the
prompt to guide the AI's response.
Chain-of-Thought - Chain-of-Thought (CoT) is a prompting technique that guides an AI to solve a problem
(CoT) through a sequence of reasoning steps.
CRISPE - CRISPE = Context + Role + Instruction + Specification + Purpose + Example
RACE - RACE = Role + Action + Context + Expectation
CLEAR - CLEAR = Context + Language + Examples + Adjustments + Refinement
The Theory of - A prompt acts as a condition that guides the AI toward a particular output.
Conditional
Generation
Prompting In GenAI - Prompting is the mathematical act of narrowing down infinite visual possibilities into a
specific subset of pixels.
Inspirational - Inspirational prompting uses loose cues that allow the model to surprise the user and is
prompting best suited for ideation.
Descriptive - Descriptive prompting uses explicit, structured instructions to achieve a specific mental
prompting image.
ReAct - ReAct is a prompting framework that combines reasoning and action, allowing an AI
system to reason about a task and take actions based on the situation.
Tree of Thoughts Tree of Thoughts is a prompting technique that explores multiple possible reasoning paths
to solve a problem.
Types of
Environments

Fully Observable - The agent's sensors give it complete access to the state of the environment at any time
(e.g., Chess).
Partially Observable - The agent has limited or noisy sensor data and cannot see everything (e.g., Self-driving
cars in traffic).
Deterministic - The current state and the agent's action completely determine the next state without
uncertainty (e.g., Tic-Tac-Toe).

Stochastic - The environment involves randomness, meaning the same action can lead to varying
outcomes (e.g., Weather or rolling dice).
Static - The environment does not change while the agent is thinking or processing (e.g., A
crossword puzzle).
Dynamic - The environment keeps changing on its own while the agent is deliberating (e.g., Driving
on a busy highway).
Discrete - There is a limited, countable number of distinct states, percepts, and actions (e.g.,
Chess).
Continuous - States and actions range over a smooth, infinite spectrum (e.g., Robotic arm steering or
fluid control).
Single-Agent - Only one intelligent agent operates alone in the system (e.g., Solving a maze).

Multi-Agent - Multiple independent or cooperative agents interact, which can be competitive (like
chess) or collaborative (like autonomous delivery fleets).
Episodic - The agent's experience is split into independent "episodes" where current actions do not
impact future choices (e.g., Batch image classification).
Sequential - Current decisions directly affect all future decisions and states (e.g., Playing a game of Go
Known - The agent fully understands the laws, rules, and outcomes of all actions.
Unknown - The agent lacks prior rules and must learn how the environment works through trial and
exploration.