| Artificial Intelligence (AI) | AI refers to the simulation of human intelligence in machines that are programmed to think, learn, and make decisions. AI can be as simple as rule-based systems or as complex as self-learning robots. The goal is to create systems that can perform tasks such as recognizing speech, making decisions, understanding natural language, and even solving problems autonomously. |
| Machine Learning (ML) | Machine Learning is a subset of AI where computers use statistical techniques to identify patterns in data and make predictions or decisions based on that data. Rather than being explicitly programmed, ML models improve over time through exposure to more data. It includes techniques like supervised learning (learning from labeled data), unsupervised learning (finding hidden patterns in unlabeled data), and reinforcement learning (learning by trial and error). |
| Natural Language Processing (NLP) | NLP is the technology behind how machines understand, interpret, and generate human (natural) language. It is used in applications like chatbots, translation systems, speech recognition, sentiment analysis, and more. NLP combines computational linguistics with machine learning to bridge the gap between human communication and computer understanding. |
| Generative AI | Generative AI focuses on creating new, original content such as text, images, audio, or code. Unlike traditional AI which often classifies or predicts based on existing data, generative AI can produce entirely new content that resembles the data it was trained on. Examples include ChatGPT (text generation), DALL·E (image generation), and Jukebox (music generation). |
| Foundation Models | These are very large AI models trained on vast amounts of diverse data (e.g., the entire internet) and designed to serve as a base for many different tasks. Once trained, foundation models can be fine-tuned with smaller, task-specific datasets to perform specialized functions. Their broad knowledge and flexibility make them a foundational layer in modern AI applications. Examples include GPT-4, BERT, and CLIP. |
| Multimodal Foundation Models | These advanced models can process and generate multiple types of data—such as text, images, and audio—at the same time. They are designed to understand complex inputs that involve more than one mode of communication (e.g., describing an image with text or answering questions about a video). This allows for more natural, human-like interaction with AI. Examples: OpenAI’s GPT-4 with vision, Google Gemini, or Meta’s ImageBind. |
| Diffusion Models | Diffusion models are a type of generative model that work by simulating a process of gradually adding noise to data and then learning to reverse that noise to recreate the original input. This reverse process can then be used to generate new, high-quality outputs such as realistic images. These models are used in image generators like Stable Diffusion and DALL·E 3. They are known for producing detailed, coherent, and high-resolution outputs. |
| Prompt Engineering | This is the practice of carefully designing and wording prompts (inputs) to elicit the best or most accurate responses from generative AI models. Since large language models can interpret many forms of input, the way a question or instruction is framed can greatly affect the outcome. Prompt engineering involves understanding the model’s behavior and experimenting with different formats, instructions, or examples to get better results. |
| Prompt Tuning | A more advanced and technical version of prompt engineering, prompt tuning involves training special prompt tokens (like placeholder text) that help guide the model’s behavior. These tokens are optimized using machine learning and are stored separately from the model, meaning you can improve task performance without retraining the entire large model. It is especially useful when resources are limited or model access is restricted. |
| Large Language Models (LLMs) | LLMs are a type of foundation model trained on massive amounts of text data (books, articles, websites) to understand and generate human language. They are capable of answering questions, writing essays, translating languages, summarizing documents, coding, and much more. LLMs use deep learning architectures (usually transformers) and contain billions of parameters, enabling them to capture complex patterns in language. Examples include OpenAI’s GPT-4, Google’s PaLM, and Meta’s LLaMA. |