SI LEARNING CENTER
AI Terms Explained in Plain English
You don't need a computer science degree to understand the language of artificial intelligence.
AI conversations can quickly become filled with abbreviations and technical terms: AI, AGI, LLM, tokens, prompts, hallucinations, agents and machine learning.
Sometimes the terminology makes the technology sound more complicated than it needs to be.
This glossary explains some of the most common terms in ordinary language.
Computer systems designed to perform tasks associated with human intelligence.
Those tasks can include understanding language, recognizing patterns, generating text and images, analyzing information, solving problems and making predictions.
AI is the broad term covering many of the intelligent technologies people are now using.
An AI system designed to perform tasks or take a series of steps toward a goal.
Instead of answering only one question, an agent may be able to plan several steps, use available tools, gather information and complete parts of a larger task.
Agents represent a shift from AI that mainly answers questions toward AI that can help carry out work.
A term generally used for a hypothetical AI with broad intellectual abilities across many different kinds of tasks.
There is no universally accepted test or definition that determines exactly when an AI system should be considered AGI.
AGI is frequently discussed when people talk about the possible future development of increasingly capable artificial intelligence.
Software designed to communicate with people through conversation.
Modern AI chatbots can answer questions, explain subjects, brainstorm ideas, help write material and assist with many other tasks through ordinary language.
Conversation has become one of the easiest ways for ordinary people to access powerful AI systems.
The background information that helps AI understand what you mean and what you are trying to accomplish.
Context might include your goal, audience, limitations, previous work or other information relevant to the request.
The more useful context AI has, the better chance it has of producing a response suited to your actual situation.
A type of machine learning that uses multi-layered neural networks to learn complex patterns from data.
Deep learning has contributed to major advances in language processing, image recognition, speech systems and generative AI.
Many modern AI capabilities grew rapidly because deep-learning systems became increasingly powerful.
AI that can generate new content in response to instructions.
Depending on the system, generative AI can produce text, images, audio, video, computer code and other material.
Generative AI changed the relationship between people and computers from mainly retrieving information to actively creating with intelligence.
An AI-generated statement that appears plausible but is inaccurate or fabricated.
An AI might provide an incorrect date, invent a source, misattribute a quotation or confidently present information that isn't true.
AI can sound confident even when it is wrong. Important information should be verified.
The combination of human experience, judgment and purpose with artificial intelligence capabilities.
Instead of viewing the future only as humans versus machines, Human + AI asks what becomes possible when the two work together.
Human + AI is one of the central ideas behind the SI — Super Intelligence project.
An AI model trained to work with and generate language.
Large language models learn statistical patterns from large amounts of text and other training data. They can generate responses based on the information and instructions provided to them.
LLMs power many of the conversational AI tools people use today.
A way of building computer systems that learn patterns from data rather than relying only on explicitly programmed rules.
A machine-learning system can use examples and data to improve its ability to classify, predict or generate results.
Machine learning is one of the major technologies underlying modern artificial intelligence.
AI that can work with more than one kind of information.
A multimodal system might work with combinations of text, images, audio, video or other forms of data.
AI is increasingly moving beyond text-only conversations toward systems capable of understanding and creating across several forms of media.
A computing model made of connected layers that can learn patterns from data.
Neural networks are loosely inspired by the idea of interconnected neurons, although they do not work exactly like a human brain.
Neural networks are an important foundation of many modern AI systems.
The instruction, question or information you give an AI.
A prompt can be as simple as:
“Explain this to me.”
Or it can include detailed instructions, examples, background information and a desired format.
The prompt begins the interaction, but you do not need to create the perfect prompt before you begin.
Learn how to work with AI without complicated prompt formulas →
The practice of designing instructions and context to help an AI produce useful results.
Prompt engineering can become technical in specialized situations, but everyday users usually do not need complicated formulas.
Clear goals and useful context often matter more to beginners than memorizing elaborate prompt templates.
Intelligence or intellectual capability beyond ordinary human ability.
Traditionally, superintelligence often refers to a hypothetical machine intelligence exceeding human capabilities across many or nearly all important intellectual areas.
The SI project also explores a broader question:
What happens when ordinary human intelligence gains access to extraordinary machine intelligence?
Perhaps the important story isn't only how intelligent machines become, but how capable humans become when extraordinary intelligence is available to them.
A term commonly used for hypothetical intelligence that significantly exceeds human intellectual capability.
Definitions vary, particularly regarding which abilities would need to exceed human performance and how broadly those abilities would need to apply.
The concept appears frequently in discussions about the possible long-term development of AI.
A small unit of text that an AI language model processes.
A token may represent a whole word, part of a word, punctuation or another small piece of text.
AI systems often measure the amount of text they can process or generate using tokens rather than simply counting words.
Information used during the process of training an AI model.
Depending on the system, training material can include text, images, audio, code and other types of data.
The patterns AI learns depend partly on the information and methods used during training.
You can understand what a tool does without knowing every technical detail about how it works.
The goal isn't to memorize AI terminology. The goal is to become comfortable enough with the language that it no longer stands between you and the intelligence.
Computer systems performing tasks associated with intelligence.
What you ask or tell the AI.
The useful background information you give the AI.
A plausible-sounding AI response that contains inaccurate or fabricated information.
Human purpose and judgment combined with machine intelligence and capability.
Knowing what AI terminology means is useful.
But the real value comes from using the technology to accomplish something.
If you're new to AI, return to the beginning of the SI Learning Center and start with something you would genuinely like to learn, solve, create or accomplish.
SI Learning Center: Learning Center | How to Work With AI | SI in Action | Using AI Responsibly | AI Readiness Quiz | Glossary