Another way to say “technology”

When an AI model confidently generates false information — citing non-existent sources, inventing people, fabricating quotes. A core limitation of LLMs.

Building software by describing what you want to an AI and accepting whatever it produces — often without reading the code. Less engineering, more directing.

Cross-Origin Resource Sharing. The browser security feature that blocks your frontend from calling APIs on a different domain unless the server explicitly allows it. Universal source of developer despair.

Low-effort AI-generated content flooding the internet — generic images, soulless blog posts, fake product reviews. The 2024–2026 equivalent of spam.

Application Programming Interface. A contract that lets one program talk to another — endpoints, parameters, responses. Modern APIs are typically HTTP + JSON.

Accumulated cost of shortcuts taken in code. Like financial debt, it compounds — the longer you wait to fix it, the harder future changes become. Metaphor coined by Ward Cunningham (inventor of the wiki) in 1992. Every codebase has it; the discipline is naming it before refactoring.

The craft of writing inputs to LLMs to reliably produce the output you want. Includes role-setting, few-shot examples, chain-of-thought, and careful phrasing.

Excessive, sycophantic praise from an AI assistant — "What a brilliant question!" style. OpenAI famously rolled back a GPT-4o update in April 2025 for being too glazing.

A call to release code to production. Prioritizes getting in front of users over chasing perfection. "Ship it" is the engineering equivalent of ripping off a band-aid — enabled by feature flags, gradual rollouts, and a willingness to fix forward rather than delay.

Minimum Viable Product. The simplest version of a product that validates a core hypothesis with real users. The purpose is learning, not launching.

Retrieval-Augmented Generation. A pattern where an LLM pulls relevant documents from a vector database before answering, grounding output in real information.

"Looks Good To Me." Shorthand approval on a pull request or code review, signaling the reviewer has no blocking concerns. Can be left solo or stacked with emojis — "LGTM 🚀" for ship-it energy.

Synthetic media — usually video or audio — where AI convincingly impersonates a real person. Went from research curiosity (2017) to mainstream policy concern (2024+). A reminder to verify things your eyes tell you are true.

A hypothetical AI that surpasses the cognitive performance of the smartest humans across virtually every domain. Distinct from AGI (human-level) and intensely debated as either imminent or fantasy.

Debugging by explaining a problem aloud to an inanimate object — traditionally a rubber duck, nowadays often an AI chatbot. Articulating the issue often reveals the solution.

The maximum amount of text an LLM can "see" at once, measured in tokens. Longer windows let models process whole codebases or novels in a single request. Claude's 1M-token context unlocked whole-repo pair programming in 2025.

A representation an AI builds of how the world works — its physics, objects, causality — that lets it predict and plan. Current LLMs have shaky world models; video models like Sora and Veo show emergent ones.

The ability of an LLM to call external functions (search, code execution, databases) mid-conversation. The building block of agentic systems.

Running a trained AI model to produce output. Training teaches the model; inference is every time you use it. Inference is what costs money per token.

Continuing to train a foundation model on a smaller, specialized dataset so it specializes in your task. Increasingly replaced by RAG + prompt engineering for most use cases.

Describing AI systems that can plan, take actions, and use tools autonomously across multi-step tasks — not just answer one prompt. The defining tech buzzword of 2025. Every vendor calls themselves agentic; half deliver a single tool call in a loop.

Derogatory slang for an AI system, robot, or automated agent — especially when it fails spectacularly. Exploded on TikTok in 2025 as a catch-all insult for AI chatbots. Has a dark-humor register that "robot" never did — casting AI as clanking and clumsy rather than sleek.

A personal knowledge system — Notion, Obsidian, Roam, Tana — where you dump notes, highlights, ideas, and re-read them as a "second brain." Tiago Forte's 2022 book codified it. Half the people with a Second Brain admit they spend more time organizing than remembering.

Large Language Model. The AI systems (GPT, Claude, Gemini) trained on massive text corpora that power modern chatbots and coding assistants. Term entered mainstream after ChatGPT's launch in November 2022; now "the AI" in most everyday speech.

The lifecycle of online platforms: first they're good to users, then they squeeze users to please advertisers, then they squeeze advertisers to please shareholders, and finally they're useless. Coined by Cory Doctorow, January 2023 WIRED.

The rate of data flow in digital networks typically measured in bits per second; the bitrate.

aparelho eletrônico, em formato de vara ou bastão, que, segurado por uma das extremidades por uma pessoa, tendo um telefone celular ou câmera fotográfica digital acoplado à outra extremidade, permite à pessoa tirar uma foto de si mesma (uma sélfie), a uma maior distância e com um melhor ângulo do que conseguiria se apenas usasse a extensão natural do próprio braço

o mesmo que pau de sélfie

To carry out a smurf attack against someone.