Skip to content
Search

Artificial intelligence

lastmod 2026-06-14created 2014-10-25135 edits

Subfields of artificial intelligence

Artificial intelligence (人工知能) philosophically means intelligence created by a being or system endowed with humanity or intellect, that is, artificial intelligence. It is generally assumed to apply to general-purpose computers. The term also refers to the field of science that studies the methodologies and feasibility of creating such intelligence.

Category

Sandbox

  • ForeverVM - A sandbox for safely running and persisting AI-generated code
  • ERA - A local sandbox for safely running AI-generated code
  • SmythOS/SRE - A cloud-native runtime for agentic AI

Projects

  • AI tool collection - AI tools worth knowing to understand the future | GeekNews
  • Runway
  • DiffusionBee
    • A program that makes Stable Diffusion available as a Mac app. Enter a text prompt and it generates an image.
    • Optimized for M1 and M2; 16 GB of RAM is reportedly recommended. It reportedly runs on 8 GB of RAM too, but took about 10 minutes.
  • LexicaArt - A site where people can post and view prompts for text-to-image models. Mostly Stable Diffusion prompts and images are posted.
  • metaphor - A search system that, given a prompt, predicts which link would come next.
  • soundraw - A site where AI generates music. Downloading music requires a paid subscription. (16 dollars yearly, 19 dollars monthly)
  • clipdrop - Lets you adjust the lighting of existing photos. It can also remove objects from an image, enlarge it (upscale), remove the background, and so on. You adjust the lighting by moving circles.
  • talk to books - Many probably know this one: enter a search term and AI finds sentences in books. Unfortunately, Korean is not supported.
  • stable diffusion hugging face - A site you can use when you just want to generate a few images with Stable Diffusion. The Stable Diffusion version 2 demo is currently available.
  • gpt-3 play ground - Good for trying out the gpt-3 language generation model.
  • Rift - An AI-Native language server that helps with development
  • Korean: 믹스보드 (Mixboard) - An experimental AI-based concept board that helps you explore and refine ideas, explore and refine ideas, and develop them.

Automation

Korean: 검색 엔진

Development (coding)

Music/voice (TTS)

Video

Design (UI/UX)

  • Tools that shorten idea → structure: Relume, Uizard
  • Tools that shorten structure → working design flow: Figma Make, Stitch
  • Tools that shorten design → actual screens (production/prototype): Framer AI, v0

Images

  • Empty Canvas - Anime-style image generation & search engine
  • ImageFX - Unlimited image generation made by Google ⭐
  • Nanobanana
  • Recraft - When you need various style samples
  • FLUX1.1 - Photorealistic images
  • FLUX.1 Kontext - Edit photos to the look you want (inpainting)

Brand Logo design

  • Looka - Looka is a site that generates logo designs with AI, making it easy to create a company icon. It offers a variety of styles and icons so you get results quickly.
  • Hatchful by Shopify - Hatchful, provided by Shopify, is a free tool that automatically creates brand icons with AI. It can generate icons suited to various industries and styles.
  • Tailor Brands - Tailor Brands is a platform that uses AI to design brand icons, logos, business cards, and more.

Pixel Art

PPT/graphs/presentations

Document processing (writing)

  • Rowfill - Provides advanced AI features for extracting and analyzing data from complex documents, images, PDFs, and more.
  • Lilys AI - Summarizes everything: video, audio, PDFs, websites ⭐
  • Korean: 클로드 (Claude)

Search/websites/web browsers

Marketing

  • Pomelli - Google's AI marketing tool

CAD

Modeling

Visualizer

  • Netron - A visualization tool for neural network, deep learning, and machine learning models.

SQL

  • data-peek - A minimal, fast desktop SQL client with AI-based SQL queries
  • Pro tip) Top 10 recommended AI tools for office workers wow.jpg | Humor board | RULIWEB
  • Claude - Document writing
    • An AI assistant you can talk with naturally in everyday conversation
    • Easily writes business documents too
    • More efficient structuring and revising of proposals and reports
  • Clova Note - Meeting minutes
    • Transcribes meetings in real time and writes summaries
    • Extracts each speaker's key points to produce concise minutes
  • Gamma - Presentation creation
    • Automatically creates easy-to-read slides from text
    • Suggests the optimal structure and design, supporting consultation materials and proposal writing
  • Notion AI - Task management
    • Organizes work easily, from task management to progress notifications
    • Team information sharing and project management in a single tool
  • Mapify - Material summarization
    • Converts long documents, materials, YouTube, or website content into an easy-to-read density
    • Makes organizing ideas and information simple, helping with thinking and organizing
  • V0 - UI component generation
    • Automatically generates UI from text prototypes, delivered as React code
    • Quickly visualizes designers' ideas, making collaboration easy
    • Generated code connects easily to Next.js projects, improving development efficiency
  • Runway - Video production
    • Automatically generates videos from text
    • Intuitive controls make it possible to produce corporate PR or social media videos
  • Ideogram - Image creation
    • AI image styling settings make it easy to generate the intended image
    • Commercial use allowed; provides AI images refined with text
  • Napkin AI - Diagram creation
    • Sketch ideas right on your knee and turn them into stories and diagrams
    • Organizes complex concepts easily, useful like notes
  • Genspark - Information gathering
    • Provides large amounts of organized related information, with analysis including up-to-date sources (URLs)
    • Can collect and integrate web page information, useful for gathering reference material

Game

  • Muse - A generative artificial intelligence (AI) model for game development made by Microsoft (MS).

Free or cheap services

A hybrid strategy that runs several free web AI models side by side, separating problem solving from code generation and playing to each model's strengths

  • Use AI Code Prep GUI to select and organize only the code you need, avoiding performance loss from unnecessary context and giving the AI only the essential context
  • Planning and debugging with high-performance/free models (Gemini 2.5 Pro, o3, o4-mini, Claude 4, etc.); execution and code writing with GPT-4.1 and Claude 3.5
  • Minimize costs by securing free or cheap tokens through OpenAI data sharing, GitHub Copilot, Poe.com, OpenRouter, etc.
  • Improve efficiency by combining various agents and CLI tools such as Claude Code, Qwen Code, Gemini CLI, Roo Code, and Trae IDE depending on the situation

My Browser Setup: The Free AI Buffet

Keeps free versions of various powerful AI models open in multiple browser tabs. Rather than relying on a single model, gets answers from multiple perspectives. The typical combination of free models used is as follows.

  • GLM 4.5: Free on the web; feels as good as Claude 4 or better. Always keeps 2–3 tabs open
  • Kimi K2: A model similar to the Claude or Opus family, free on the web. Usually 1–2 tabs open; before GLM 4.5 appeared, it solved tricky bugs several times a day
  • Qwen3 Coder and newer models: Used to test various coding-specialized models
  • OpenAI Playground: Various models such as GPT-4.5 and o3 available for free. Allowing "OpenAI to use data for model training" in the account data settings grants free tokens
  • Google Gemini AI Studio: Gemini 2.5 Pro/Flash models free and nearly unlimited. Usually 1–3 tabs open
  • Google Gemini 2.5 Pro: A separate service from AI Studio, better at image generation and deep research. Used in tabs alongside AI Studio
  • Poe.com: Daily free credits for premium models such as Claude 4 and o4-mini
  • OpenRouter: A mix of free and paid models. Several models set up in separate tabs
  • ChatGPT: The free version is still useful; keeps at least 1 tab
  • Perplexity AI: Strong at research-oriented questions
  • Deepseek: v3 and r1 models free. Watch out for context limits, though
  • Grok.com: Unlimited free for general use, deep research, and image editing. The deep research feature in particular is useful, similar to Perplexity
  • Phind: Tries to include flowcharts or diagrams with answers
  • lmarena.ai: Claude Opus 4 and Sonnet 4 for free. Free use of Opus 4 is quite valuable

Terms

[Recommended] "2025 Key glossary for the AI era, explored through AI trends and issues" published by NIA (National Information Society Agency)
Key glossary you must know in the AI era (168p PDF) | GeekNews
https://www.nia.or.kr/site/nia_kor/ex/bbs/View.do?cbIdx=82618&bcIdx=28928&parentSeq=28928
Nia.or.kr_-ai-_2025_terms.pdf
  • Affective computing / Reinforcement learning / Large language model (LLM) / Large action model (LAM) / Retrieval-augmented generation (RAG)
  • Small language model (SLM) / High bandwidth memory (HBM) / Spatial intelligence / Overfitting / Optical character recognition (OCR)
  • Graphics processing unit (GPU) / Symbolic AI / Neuromorphic computing / Conversational AI / Data labeling
  • Data wrangling / Data silo / Data preprocessing / Data flywheel / Deep learning
  • Deepfake / Low code / Parameter / Machine learning / Machine learning operations (MLOps)
  • Multimodal / Processing-in-memory (PIM) / Metadata / Meta-learning / Model compression
  • Model Context Protocol (MCP) / Fine-tuning / Vibe coding / Vertical AI / Benchmark dataset
  • Distributed training / Vision language model (VLM) / Chain of thought (CoT) / Pretrained model / Generative adversarial network (GAN)
  • Generative AI / Shadow AI / AI as a service (AIaaS) / Explainable AI (XAI) / Recurrent neural network (RNN)
  • Simulation-to-reality transfer (Sim-to-Real) / Autoencoder / Open-source AI / On-device AI / World model
  • Anomaly detection / Causal AI / Embedding / Automated machine learning (AutoML) / Natural language processing
  • Low-rank adaptation (LoRA) / Accuracy / Zero-shot learning / Intelligent base station (AI-RAN) / Artificial intelligence of things (AIoT)
  • Supervised learning / Knowledge distillation / Curse of dimensionality / Test-time compute (TTC) / Jailbreak
  • Token / Turing test / Transformer architecture / Foundation model / Discriminative AI
  • Fabless / Frontier AI / Prompt / Prompt injection / Physical AI
  • FinFET / Convolutional neural network (CNN) / Synthetic data / Hallucination / AI guardrails
  • AI accelerator / AI divide / AI native / AI data center / AI red teaming
  • AI literacy / AI semiconductor / AI trustworthiness / AI safety / AI assistant
  • AI agent / AI orchestration / AI watermarking / AI ethics / AI transformation (AX)
  • AI alignment / AI reasoning (Reasoning) / AI inference (Inference) / AI bias / AI human

How consumer psychology affects AI product design

See Product#How consumer psychology affects AI product design.

A practical guide to improving AI products quickly

See Product#A practical guide to improving AI products quickly.

Beyond the AI MVP: What you actually need

See Intelligent agent#Beyond the AI MVP: What you actually need.

Assistive agent optimization (AAO)

A four-stage evolution: SEO -> AEO -> AIEO -> AAO:

  1. Search engine optimization (SEO) — be found.
  2. Answer engine optimization (AEO) — be the answer.
  3. AI engine optimization (AIEO) — be the recommendation.
  4. Assistive agent optimization (AAO) — be chosen when no human is in the loop.

The announced 10-stage DSCRI-ARGDW pipeline

  • DSCRI-ARGDW, to be published later, is the 10-stage process by which content leads to a conversion through an AI engine
    • Discovered: existence recognized
    • Selected: judged worth collecting
    • Crawled: content collected
    • Rendered: converted into a readable form
    • Indexed: stored in the algorithm
    • Annotated: semantically classified
    • Recruited: selected for use
    • Grounded: trust verified
    • Displayed: shown to the user
    • Won: wins the AI's single recommendation (the 'perfect click')
  • This pipeline describes the entire journey of content being selected by an AI agent and leading to a conversion

Microsoft's introductory Edge AI course

This course is designed to guide beginners through the exciting world of Edge AI, covering fundamental concepts, popular models, inference techniques, device-specific applications, model optimization, and the development of intelligent Edge AI agents.

  • A free, open-source course for learning step by step how to run AI models directly on edge devices
  • The goal is to run AI models on the device rather than in the cloud, achieving minimal latency, privacy, cost savings, and less network dependence
  • Learners complete hands-on projects such as SLM deployment, building multi-agent systems, implementing local RAG pipelines, and model routing, and learn deployment for Windows, mobile, embedded, and hybrid environments
  • Explains, with a hands-on focus, how to improve performance (up to 85% faster, 75% smaller models) with key optimization tools such as Llama.cpp, Microsoft Olive, OpenVINO, and Apple MLX

Curriculum (total estimated time: 36–45 hours)

  1. EdgeAI overview and learning objectives (Introductory: 1–2 hours)
  2. Edge AI fundamentals and case studies (Beginner: 3–4 hours)
  3. SLM model architecture and families (Beginner: 4–5 hours)
  4. Local and cloud deployment practice (Intermediate: 4–5 hours)
  5. Model optimization toolkits (Llama.cpp, etc.) (Intermediate: 5–6 hours)
  6. SLMOps and fine-tuning (Advanced: 5–6 hours)
  7. AI agents and Function Calling (Advanced: 4–5 hours)
  8. Cross-platform implementation (Advanced: 3–4 hours)
  9. Foundry Local sample projects (Expert: 8–10 hours)

From generating a song with AI to releasing it on Spotify

Workflow:

  • Prepare the song form: Intro-Verse 1->Pre-Chorus->Chorus(Hook)->Verse 2 .. etc.
  • Write lyrics that fit the song form, and adapt them into the desired language with an LLM
  • Generate the song in Suno with prompts for lyrics/genre/sound, etc. (the prompts themselves are also discussed with an LLM)
  • Generate several times, pick one, then post-production such as mastering
  • Apply for music distribution: RouteNote (free, 15% fee on music revenue); there are also DistroKid/Amuse (paid), etc.
  • Cover: draft with Midjourney -> i2i with Stable Diffusion -> add fonts in Canva

See also

Favorite site

author Zer0 sitemap