Charts
DataOn-chain
VIP
Market Cap
API
Rankings
CoinOSNew
Language
  • 简体中文
  • 繁体中文
  • English
Leader in global market data applications, committed to providing valuable information more efficiently.

Features

  • Real-time Data
  • Special Features
  • AI Grid

Services

  • News
  • Open Data(API)
  • Institutional Services

Downloads

  • Desktop
  • Android
  • iOS

Contact Us

  • Chat Room
  • Business Email
  • Official Email
  • Official Verification

Join Community

  • Telegram
  • Twitter
  • Discord

© Copyright 2013-2026. All rights reserved.

简体繁體English
|Legacy

Cryptocurrency Circle People Learn About AI: 30 High-Frequency Jargon Terms Explained All at Once

CN
律动BlockBeats
Follow
3 hours ago
AI summarizes in 5 seconds.
Original title: "AI Jargon Dictionary (March 2026 Edition), Recommended for Collection"
Original author: Golem, Odaily Planet Daily

Now, if people in the crypto circle do not pay attention to AI, they are likely to be laughed at (yes, my friend, think about why you clicked in).

Are you completely clueless about the basic concepts of AI, asking what every acronym in each sentence means? Are you also confused by various technical terms at offline AI events but pretend not to be offline?

While it's unrealistic to jump into the AI industry in a short time, knowing some high-frequency basic vocabulary in the AI industry is definitely worthwhile. Luckily, the following article is prepared for you ↓ I sincerely suggest you read and save it.

Basic Vocabulary (12)

· LLM (Large Language Model)

The core of LLM is a deep learning model trained on massive amounts of data, proficient in understanding and generating language; it can process text and is increasingly capable of handling other types of content.

In contrast, SLM (Small Language Model) usually emphasizes lower costs, lighter deployment, and more convenient localization.

· AI Agent

An AI Agent refers not only to a "chatting model," but a system that can understand goals, invoke tools, execute tasks step-by-step, and plan and validate when necessary. Google defines an agent as software that can perform actions on behalf of users based on multimodal input.

· Multimodal

This AI model can process not just text, but also multiple input and output forms such as text, images, audio, and video simultaneously. Google explicitly defines multimodal as the ability to process and generate different types of content.

· Prompt

The instructions given by users to the model; this is the most basic mode of human-computer interaction.

· Generative AI (AIGC)

This emphasizes AI "generation" rather than simple classification or prediction. Generative models can create text, code, images, memes, videos, etc., based on prompts.

· Token

This is one of the concepts in the AI world that resembles a "Gas unit." Models do not understand content by "word count," but by processing inputs and outputs by tokens; billing, context length, and response speed are usually strongly related to tokens.

· Context Window

This refers to the total number of tokens that the model can "see" and utilize at one time, also known as the number of tokens considered or "remembered" by the model during a single processing session.

· Memory

Allows the model or agent to retain user preferences, task contexts, and historical states.

· Training

The process by which a model learns parameters from data.

· Inference

This refers to the process when the model is deployed, receiving inputs and generating outputs. It is often said in the industry that "training is expensive, but inference costs even more," because many costs occur during the commercial stage of inference. The distinction between training and inference is also the foundational framework for mainstream manufacturers when discussing deployment costs.

· Tool Use / Tool Calling

This means that the model outputs not just text but can also invoke tools like searches, code execution, databases, external APIs, etc.; this has been recognized as one of the key capabilities of an Agent.

· API

The infrastructure when AI products, applications, and agents connect to third-party services.

Advanced Vocabulary (18)

· Transformer

A model architecture that allows AI to better understand contextual relationships; it is also the technical foundation of most large language models today, characterized by its ability to consider the relationship of each word to other words in the entire content simultaneously.

· Attention

This is the most critical core mechanism of Transformer, which allows the model to automatically determine "which words are most worth focusing on" when reading a sentence.

· Agentic / Agentic Workflow

This is a recently popular term, meaning a system that is no longer just a "question-and-answer" format, but autonomously breaks down tasks, decides the next steps, and invokes external capabilities. Many manufacturers view it as the symbol of "shifting from Chatbot to executable systems."

· Subagents

A single Agent can create multiple dedicated small Agents to handle sub-tasks.

· Skills

With the explosion of OpenClaw, this term has become noticeably more common; it refers to installable, reusable, and combinable capability units/operation manuals for AI Agents, but it also specifically warns of the risks of tool misuse and data exposure.

· Hallucination

This refers to the model making nonsensical claims, "perceiving patterns that do not exist," resulting in erroneous or absurd outputs, which is an overly confident output that seems reasonable but is actually incorrect.

· Latency

The time taken from the model receiving a request to outputting results; this is one of the most common engineering jargon, frequently appearing in discussions about implementation and productization.

· Guardrails

Used to limit what the model/agent can do, when to stop, and what content cannot be output.

· Vibe Coding

This term is currently one of the hottest AI jargons, meaning users directly express their needs through conversation, and AI writes code without requiring users to specifically understand how to code.

· Parameters

The numerical scale used internally by the model to store capabilities and knowledge; it is commonly used to roughly measure model size; "billions of parameters" and "hundreds of billions of parameters" are some of the most common intimidating phrases in the AI field.

· Reasoning Model

This usually refers to models that are better at multi-step reasoning, planning, validation, and executing complex tasks.

· MCP (Model Context Protocol)

This has become a very popular new jargon in the past year, serving to establish a universal interface between the model and external tools/data sources.

· Fine-tuning / Tuning

This refers to continuing training on a base model, making it better suited for specific tasks, styles, or domains. Google's glossary directly considers tuning and fine-tuning as related concepts.

· Distillation

This involves compressing the capabilities of a large model into a smaller model, like a "teacher" teaching a "student."

· RAG (Retrieval-Augmented Generation)

This has almost become a basic configuration for enterprise AI. Microsoft defines it as a "search + LLM" model that uses external data to ground responses, addressing the issues of outdated training data and lack of knowledge of proprietary knowledge bases. The goal is to base answers on real documents and private knowledge rather than relying solely on the model's memory.

· Grounding

This often appears alongside RAG, meaning grounding model responses on external evidence such as documents, databases, and web pages, rather than relying solely on parameter-based "free play." Microsoft explicitly identifies grounding as a core value in RAG documentation.

· Embedding

This refers to encoding text, images, audio, and other content into high-dimensional numerical vectors for semantic similarity calculations.

· Benchmark

This is a standardized way of testing model capabilities and is also the language in which different models love to "prove how strong they are" in rankings.

Original link

免责声明:本文章仅代表作者个人观点,不代表本平台的立场和观点。本文章仅供信息分享,不构成对任何人的任何投资建议。用户与作者之间的任何争议,与本平台无关。如网页中刊载的文章或图片涉及侵权,请提供相关的权利证明和身份证明发送邮件到support@aicoin.com,本平台相关工作人员将会进行核查。

返20%!OKX钱包龙虾技能,AI一键自动赚
广告
|
|
APP
Windows
Mac
Share To

X

Telegram

Facebook

Reddit

CopyLink

|
|
APP
Windows
Mac
Share To

X

Telegram

Facebook

Reddit

CopyLink

Selected Articles by 律动BlockBeats

2 hours ago
Agents can also engage in mutual promotion in business; this AI hackathon by Circle was explosive.
3 hours ago
Firecrawl launched an Agent-specific web scraping tool, Nvidia released Nemotron 3 Super, what is the English community discussing today?
19 hours ago
Inherent benefits | With over 500 people registered for the event, how else can this lobster debate大会 be played?
View More

Table of Contents

|
|
APP
Windows
Mac
Share To

X

Telegram

Facebook

Reddit

CopyLink

Related Articles

avatar
avatarTechub News
20 minutes ago
Exclusive Tax Perspective: The Compliance Code and Tax Burden Logic Behind Animoca Brands' Private Equity Tokenization
avatar
avatarTechub News
1 hour ago
The story behind Circle's stock price doubling in one month.
avatar
avatarTechub News
1 hour ago
The Strait of Hormuz triggers an oil war, will the Federal Reserve bow down by lowering interest rates in June?
avatar
avatarTechub News
1 hour ago
Set sail! Envision the future · AI-empowered digital assets × Innovative management Global Leadership Camp officially ignited in Hong Kong.
avatar
avatarAiCoin
1 hour ago
2.5% trap! The Middle East conflict is rewriting the Federal Reserve's script!
APP
Windows
Mac

X

Telegram

Facebook

Reddit

CopyLink