Get Quartz in your inbox

Free daily briefing on global business news.

New AI slang, explained: 10 terms from vibe coding to clanker

This AI slang guide explains 10 terms, from vibe coding and AI slop to p(doom) and e/acc, and what each one actually means

The vocabulary of artificial intelligence no longer stays inside research labs. Words that began in academic papers, developer forums and posts on X $TWTR 0.00% now turn up in office chats, group texts and product pitches. Many of them carry an attitude along with a definition. Learning these terms is a quick way to see how people feel about the technology.

This selection of AI slang covers 10 words and phrases. They range from technical shorthand to outright insults to the language of competing ideologies. Some describe new ways of working. "Vibe coding" names a style of software development in which a person explains a goal in everyday language and lets a model handle the programming. "Agentic" describes systems that take several steps toward a goal on their own, rather than answering one prompt and stopping.

Other terms express doubt. "Stochastic parrot" comes from a 2021 academic paper that cast large language models as stitching together word patterns without grasping meaning. "Wrapper" is a mild put-down for a product built mostly on top of another company's model. "Hallucination" names a persistent problem with chatbots: confident answers that turn out to be false.

A third group is about mockery. "AI slop" describes the low-effort text, images, audio and video that generative tools produce in bulk. "Clanker," a robot insult borrowed from Star Wars, became a way for social media users to jeer at chatbots in 2025.

The last group maps the argument over where AI is headed. "P(doom)" is shorthand for a person's own guess at the odds that AI ends in catastrophe for humanity. "E/acc," short for effective accelerationism, labels the camp that wants development to speed up. "Jailbreak" sits between the technical and the political. It describes prompts that push a model past its safety rules.

None of this AI jargon is fixed. Meanings shift as the tools change, and some terms fade as fast as they arrive. Each slide that follows defines one term and, where the history is clear, explains where it came from. It also shows how people use the word now. Read together, these AI buzzwords offer a working map of a technology that now shows up in how people write, code and argue online.

Vibe coding: what it means to let AI write the code

Vibe coding is a way of building software by describing what you want to an AI model in plain language. The person then accepts the code it writes without reading it closely, steering by results rather than by reviewing each line.

Andrej Karpathy introduced the term in a post on X $TWTR 0.00% in early February 2025. Karpathy is a co-founder of OpenAI and the former head of AI at Tesla $TSLA -0.80%. The phrase has since spread well beyond software engineering circles.

The method rests on a change in who does the typing. In a vibe coding session, the human explains a goal, such as a simple web page or a small tool that sorts files. The AI model writes the code. The human runs it, sees what happens and asks for changes. When an error appears, the usual move is to paste the error message back into the chat and let the model try a fix.

That habit separates vibe coding from ordinary AI-assisted programming. Many professional developers use AI tools to suggest code, but they still read, test and edit what those tools produce. They treat the output as a draft from a fast but fallible assistant. Vibe coding drops the review step. The code is judged mainly on whether the finished program appears to work.

The approach lowers the barrier to making software. People with little or no programming training can produce a working prototype by describing it. Experienced engineers can also use it to try out an idea quickly before deciding whether it deserves a careful build.

The trade-offs come from skipping review. Code that nobody has read can contain bugs that surface later, under conditions the first tests never covered. It can also hide security gaps, such as weak handling of passwords or user data, that a working demo will not reveal. Fixing problems gets harder when the person in charge does not understand how the program is structured.

For that reason, the term often describes a kind of project as much as a technique. It is commonly applied to throwaway experiments, weekend projects and early prototypes. Software that handles money, health records or other sensitive information is a different case. Those projects generally call for code that someone can read and explain line by line.

AI slop: the term for low-quality content mass-produced by AI

AI slop is low-quality text, images, audio or video that generative AI tools produce in bulk and that gets posted online with little human care. The word describes output made to fill space or catch attention, not to inform anyone.

The label borrows from an older meaning. In everyday English, slop refers to watery, unappetizing food or the scraps fed to pigs. Applied to the internet, it suggests content that is cheap and abundant. It also suggests content made without regard for whoever ends up consuming it.

The term stretches across every format generative tools can produce. That includes articles and product reviews written by chatbots. It includes pictures from image generators that look polished at a glance but fall apart on closer inspection. It also covers synthetic voiceovers and short videos assembled from AI-generated clips. What ties these together is not the technology used but the lack of effort, review and purpose behind the result.

Low cost drives much of the volume. One person with text and image generators can produce in minutes what once took hours of human work. When posts earn money through advertising or platform payouts based on views, flooding a feed becomes a workable strategy. That holds even if most of the output is ignored.

Readers tend to meet slop in a few places. Social media feeds surface AI-generated images designed to draw likes and shares. Search results can lead to content farms full of generic text........

© Quartz