AI Vocab

AI Vocab Training

Transcript

1. AI Literacy Vocab & Terms An overview of need-to-know terms
2. As research has progressed, different types of AI, with differing capabilities, have emerged. Artificial Intelligence (AI) is a general term that refers to computer systems that can perform tasks that would previously have required human intelligence. What is AI?
3. AI can be categorized in a few different ways: By capability: Artificial Narrow Intelligence Artificial General Intelligence * Artificial Super Intelligence* By Functionallity: Reactive Machines Limited Memory Theory of Mind* Self-Aware AI* By Technology: Artificial Intelligence Machine Learning Deep Learning Natural Language Processing *Note: These types of AI are currently hypothetical (at time of writing). What types of AI are there?
4. AI Categories by Technology, A Closer Look Categories of AI by Technology: Machine Learning Deep Learning Natural Language Processing Unlike the other two categorization systems, the types of AI under this category are all currently real, and not hypothetical.
5. Machine Learning (ML) is a core branch of AI that enables computers to learn from data and improve their performance on a task without being explicitly programmed for every scenario. There are different approaches to Machine Learning, which vary depending on how the algorithm learns from data and what kind of feedback it gets. What is Machine Learning? E-mail spam filters are an example of Machine Learning.
6. Deep Learning (DL) is a subset of machine learning that uses artificial neural networks with many layers (hence "deep") to model and solve complex problems. Types of DL architecture include (but are not limited to) ; convolutional neural networks (CNN), Transformers, and Graph Neural Networks (GNNs). What is Deep Learning?
7. Natural Language Processing (NLP) is, as the name suggests, the field of AI that focuses on enabling machines to understand, interpret, and respond using human language such as English. NLP combines linguistics with machine learning and allows communication between humans and machines to feel more natural for the human. What is Natural Language Processing?
8. An AI Agent is a type of AI that can perceive its environment, make decisions, and take actions to achieve specific goals. Example: An AI agent can navigate to a travel site and book a flight. What is an AI Agent?
9. A local model stores and processes data on your own computers or servers. Local Model A cloud model stores and processes data in the cloud, i.e, on someone else’s system. Higher risk of FERPA/ HIPPA violations. Cloud Model What is a Local Model vs a Cloud Model?
10. The layer of software that surrounds the actual AI model. It facilitates interactions between users and the underlying AI model, as well as interactions between the AI model and software tools the AI has access to. It also handles the AI’s memory, system prompts, and post-processing of the AI’s output. To use a car as an analogy; think of the AI model as the engine of the car, and the wrapper as the dashboard and body. What is a Wrapper?
11. A system prompt is additional text that the wrapper invisibly adds to every prompt the AI is given, usually at the start. This is used to adjust formatting, style, logic, and modalities. Here’s an example you might see if you looked at the code, the first line being the system prompt, and the second line being the user’s prompt. What is a System Prompt? [ {system : “You are a helpful travel assistant.”}, {user: “Plan me a trip to Japan.”} ]
12. A large language model (LLM) is a type of machine learning model designed to understand, generate, and predict human language. These are what allow AIs to understand natural human language. What is a Large Language Model?
13. Transformers have been called “the core invention underlying the current boom in AI.” A transformer is a type of AI architecture, but what is an architecture? In AI, an architecture is the conceptual design and structure of a model, which determines how data gets processed into outputs. This is where lots of number crunching happens. What makes a Transformer unique is it’s use of multiple layers, each with a self-attention mechanism and a multilayer perceptron (those are different fancy math algorithms, which we won’t get into here). What is a Transformer?
14. What is a Token? Tokens are how transformers understand input. When data is inputted into a transformer, it is first broken down into little pieces called “tokens.” For example, a sentence might get broken down into words, pieces or words and punctuation marks. Next each token is represented as an array of numbers. This is called encoding.
15. https://ischool.syracuse.edu/types-of-ai/ https://www.ibm.com/think/topics/artificial-intelligence-types https://cloud.google.com/learn/what-is-artificial-intelligence Sources: