Answer:
Generative AI is a branch of Artificial Intelligence that creates new content such as text, images, audio, video, and code based on patterns learned from existing data. Examples include ChatGPT, Google Gemini, Claude, Midjourney, and DALL·E.
| Traditional AI | Generative AI |
|---|---|
| Analyzes existing data | Creates new content |
| Performs prediction/classification | Generates text, images, videos, code |
| Rule-based or ML models | Large Language Models (LLMs) and Foundation Models |
| Example: Spam Detection | Example: ChatGPT |
Answer:
A Large Language Model (LLM) is an AI model trained on massive amounts of text data to understand and generate human-like language.
Examples:
Answer:
A prompt is the instruction or input given to an AI model to generate a response.
Example:
“Write a Python program to reverse a string.”
The better the prompt, the better the output.
Answer:
Prompt Engineering is the process of designing effective prompts to get accurate and relevant responses from AI models.
Techniques include:
Answer:
Foundation Models are large pre-trained AI models trained on vast datasets that can perform multiple tasks with minimal fine-tuning.
Examples:
Answer:
GPT (Generative Pre-trained Transformer) is a family of language models developed by OpenAI.
It works in two stages:
GPT can generate:
Answer:
A Transformer is a deep learning architecture introduced by Google in 2017.
It uses:
Transformers power modern LLMs like GPT and BERT.
Answer:
Tokenization is the process of splitting text into smaller units called tokens.
Example:
Sentence:
I love AITokens:
["I", "love", "AI"]LLMs process tokens instead of full sentences.
Answer:
Tokens are the smallest units processed by an LLM.
A token may be:
Example:
Artificial Intelligencemay become
Artificial
Intelligenceor even smaller subwords depending on the tokenizer.
Answer:
Temperature controls the randomness of AI responses.
Lower temperature is ideal for coding and factual answers.
Answer:
Top-p (Nucleus Sampling) selects the next word from the smallest set of tokens whose cumulative probability exceeds p.
Example:
It balances creativity and accuracy.
Answer:
Hallucination occurs when an AI model generates false, misleading, or fabricated information while presenting it confidently.
Ways to reduce hallucinations:
Answer:
Fine-Tuning is the process of training a pre-trained model on domain-specific data to improve performance for a particular task.
Examples:
Answer:
RAG combines an LLM with an external knowledge source.
Process:
Benefits:
Answer:
Embeddings are numerical vector representations of text, images, or other data that capture semantic meaning.
Uses:
Answer:
A Vector Database stores embeddings and enables similarity search.
Popular Vector Databases:
They are widely used in RAG applications.
Answer:
LangChain is an open-source framework for building applications powered by LLMs.
Features:
Answer:
An AI Agent is an autonomous system that can:
Examples:
Answer:
| Tool | Purpose |
|---|---|
| ChatGPT | AI Chatbot |
| Google Gemini | LLM |
| Claude | AI Assistant |
| Microsoft Copilot | Coding & Productivity |
| GitHub Copilot | Code Generation |
| DALL·E | Image Generation |
| Midjourney | AI Art |
| Stable Diffusion | Image Generation |
| Runway ML | AI Video Generation |
| LangChain | LLM Application Framework |
| LlamaIndex | RAG Framework |
| Hugging Face | AI Models & Libraries |