In the field of AI, the term “Vibe” has suddenly exploded, with concepts ranging from “Vibe Coding” to “Vibe Marketing” emerging one after another. It lowers the technical threshold and allows the public to enjoy AI dividends, but it has also been criticized as over-packaging “black words” that lack substance. This article delves into the process of “Vibe” from its origins to widespread dissemination, as well as the controversies and reflections it has sparked in different fields.
How many times have you heard Vibe in the past month?
If you pay attention to various developments in AI, you should hear more than you have heard in the past N years.
From Vibe Coding, to Vibe Marketing, Vibe Investing, …… “Vibe” is becoming a kind of Vibe in the AI circle.
Some people regard it as a gospel for the popularization of technology, so that ordinary people can also share the dividends of AI; Some people also think that this is a new round of concept packaging, which is turning all products into a kind of “small circle black talk” with nothing to say.
Some people are crazy about “Vibe+ everything”, firmly believing that this represents the next era of human-computer interaction; Some people have had a complete fed up with the word flying all over the sky, fearing that it is hollowing out the core of the technical discussion.
Why did the word “Vibe” become the subject of collective discussion?
Karpathy’s tweet, Vibe became a hot word
It all started in February 2025 with a tweet on X by Andrej Karpathy, a founding team member of OpenAI and former Tesla AI director.
What does a product manager need to do?
In the process of a product from scratch, it is not easy to do a good job in the role of product manager, in addition to the well-known writing requirements, writing requirements, writing requirements, there are many things to do. The product manager is not what you think, but will only ask you for trouble, make a request:
View details >
In it, he describes a new programming experience called “Vibe Coding”, “I was completely immersed in the atmosphere, enjoying the exponential explosion of productivity, and forgetting about the code. By chatting with AI, you can directly complete tasks such as “reducing the margin of the sidebar by half” and “accept” the AI’s operations.
If the code is reported as an error, just copy and paste the error message and throw it to AI to regenerate, “It feels like I’m not writing code, I’m just watching, talking, running, copying and pasting, and then it’s done.” He said.
The great thing about Karpathy is that he didn’t make any cold technical slang, but chose the word “Vibe”.
The traditional term “AI Coding” has its own limitations – it sounds too technical and is aimed at developers who already have a programming foundation. The programmer group has never been a “sexy” existence in popular culture.
But when the label “Vibe” was added before coding, everything changed. “Vibe” suggests that programming is no longer a patent of a technocratic party with a threshold, but a sensual, intuitive act of creation. This expression instantly shortens the distance between technology and the public.
The tweet finally exceeded 5,000 retweets and more than 20,000 likes, like throwing a spark on dry firewood, accurately capturing the collective sentiment of an era, and then many practitioners in the circle and major media began to widely use this word.
The tech circle has also found the perfect spiritual totem for this movement – legendary music producer Rick Rubin, as the winner of nine Grammys, Rubin has admitted in countless interviews that he “hardly knows how to play musical instruments or operate a mixer”, and he relies on his intuition for good voices and his grasp of the “feeling” of the work.
A grandmaster with “zero technology” and pure taste, but can create countless industry classics with “Vibe” – isn’t this the most perfect humanoid incarnation of “Vibe Coding”?
A promotional photo taken at the 2016 CES exhibition was dug up, in which Rick Rubin wears headphones with his eyes closed, looking like he is “feeling with his soul”, which has become the perfect meme for virality.
Even, some time ago, Rick Rubin really collaborated with Anthropic to publish a book called “The Way of Code”, which was actually created by “AI code master” Claude, transforming the core ideas of the Tao Te Ching into the way of code. Taoism’s “the road is invisible, governs without doing anything, and goes with the flow” and Vibe Coding’s “regardless of the process, follow AI, and let technology take you fly” is abstractly combined.
A Vibe Coder seems to only need to write prompts to let AI do everything.
From Vibe Coding to Vibe Everything
If Karpathy ignited the fuse, then what happened next was a chain explosion.
The marketing circle quickly borrowed it. Flowith’s CMO kidnapper mentioned that “Vibe Marketing” represents a new marketing method – abandoning the cumbersome research, analysis, and planning processes of traditional marketing and relying on the unlimited firepower and rapid verification of AI.
From Abductor:
1. Don’t reject AI generation: Instead of marketing teams polishing every ad word, let AI generate a large number of copy, images, video scripts, and even user portraits and marketing strategies.
2. Choose and iterate based on “feeling”: In the face of the massive options provided by AI, the decision is simplified to let the AI make decisions “this feeling is right” and “this looks like a hit”. Instead of delving into the logical chain, it is quickly launched for testing.
3. Quick response rather than in-depth planning: When receiving market feedback, the action is often “copy and paste” adjustments – quickly changing assets, adjusting bids, trying new AI suggestions, and not delving into the root cause.
4. The “Schrödinger state” of the strategy: Marketing activities may go beyond the rigorous framework of traditional marketing theory or even deviate from the brand’s long-term strategy. Speed and “net feel” sometimes overwhelm brand consistency.
5. Chasing tools and trends: Marketers are always catching up with the latest AI tools, platform algorithms, or social media hotspots, knowing that these “best practices” can be fleeting.
“On the one hand, Vibe Marketing gives you unprecedented efficiency magic, but on the other hand, it also makes you dizzy in the tool chain of information and change, feeling like you’re constantly adjusting and testing, but you’re spinning in place, creating more ‘marketing noise’ than clear brand value.” He added, “Traditional marketing pursues precision, while Vibe Marketing pursues ‘resonance’. In today’s information overload, users won’t remember your entire argument, but they will remember how you made them feel. ”
The same is true in the field of design. Traditional UI/UX design requires designers to master color theory, typography principles, interaction logic, and other expertise, but “Vibe Design” tries to replace all of this with a more intuitive way.
Platforms like Lovart allow users to generate interface designs by describing “I want that quiet, spa-like feel” or “a bit of cyberpunk coolness” without having to know any technical details such as hex color values, font hierarchies, and responsive layouts. As long as the user can express the desired “feeling”, AI can transform this vague emotional need into a concrete visual presentation.
In the field of content creation, YouMind founder Yu Bo mentioned the concept of “Vibe Writing” in an interview, which he believes represents a paradigm shift in content creation. “When creation is no longer about publishing papers or writing professional reports, but about sharing a moment of inspiration with friends, or recording the feelings of a trip, AI can greatly ‘lower the psychological threshold for creation’.”
Some entrepreneurs with a keen sense of smell have begun to quietly change their profiles in social media profiles, from “Senior Developer” to “Vibe Coder”, from “UX Designer” to “Vibe Designer”, this repackaging of professional identity also reflects the subtle changes of the entire industry.
Quickly turned into a “black word” for self-entertainment?
In our article yesterday, he “invented” “Vibe editing”, and said that according to this circle, he should call himself “Vibe Editing Proposer Definer Concept Manager”.
In addition to wanting to “smile” with our readers, we do feel that this word is starting to move towards the final direction of many concepts in the AI industry today: there is no clear definition, anyone can use it.
When any concept is overused, over-interpreted, over-packaged, a backlash is almost inevitable. Reflections on the word “Vibe” have also begun.
The earliest criticism came from within the tech community. A senior developer posted an article titled “The Hidden Costs of Vibe Coding”, one of which was reposted wildly: “Congratulations, now two engineers can produce the same amount of insecure, unmaintainable, and technically indebted code that only 50 engineers could write. ”
Although this is obviously ironic, it hits the core question: when we rely too much on AI to generate code, are we improving efficiency or accumulating risks? In the article, Delbare analyzed several “Vibe Coding” project cases in detail and found that although these projects were developed quickly, they had serious hidden dangers in long-term maintenance, performance optimization, and security audits.
At an AI conference in May 2025, Andrew Ng described “Vibe Coding” as a “dangerous misleading concept.”
“People think that ‘Vibe Coding’ means writing code without using their brains, which is completely wrong,” Ng mentioned, “In fact, effectively guiding AI to write useful software is an extremely deep intellectual exercise that requires a lot of thinking, supervision and professional judgment.” When I spent the whole day programming with the help of AI, to be honest, I was already too tired at night. ”
What’s even more interesting is that even Karpathy, the creator of “Vibe Coding”, began to revise his concept. In recent public speeches, he has started pushing a new concept: “AI-assisted coding.”
“A friend asked me a few days ago if I was coding in vibe and I said no, I was coding in real,” Karpathy tweeted. In his view, truly effective human-computer collaborative programming requires developers to maintain professional technical judgment and a deep understanding of code logic and architecture, rather than blindly relying on “feeling”.
This shift is subtle but significant.
In China, Hong Dingkun, vice president of technology at ByteDance, proposed a more differentiated concept at the Volcano Engine Conference: “Write code in natural language”. He believes that this is fundamentally different from “Vibe Coding”.
“Leaving everything to AI, only responsible for clicking buttons and raising requirements, the program produced in this way is difficult to maintain,” Hong Dingkun said, “Writing code in natural language is not to vaguely describe the function and then wait for the result, but to use natural language to accurately describe the coding logic and technical solutions.” ”
Hong Dingkun shared his experience of using Trea: “When I propose technical solutions and processes, I will be very specific, and I will carefully review the generated code, so that I can take over manual modifications at any time.” ”
From the perspective of different AI programming products on Vibe Coding, we can also see how they position themselves, user groups and product routes. Some products choose to embrace the “Vibe” concept, emphasizing ease of use and lowering barriers; Others choose to distance themselves from “Vibe”, emphasizing professionalism and code quality.
When Sequoia Capital used the term “Vibe Revenue” in a recent industry report, there was also some irony in the joke. This concept is specifically used to describe early revenues that are primarily driven by market hype, user curiosity, and concept popularity, rather than real value creation.
Let Vibe belong to Vibe
The popularity of a word is the most accurate slice of the mood of the times. In the field of artificial intelligence, which is based on precision, logic and computing power, why has a vague, emotional and even somewhat frivolous word like “Vibe” become the new lingua franca?
Perhaps the answer lies in the fact that “Vibe” perfectly fills the narrative vacuum of current AI development.
Artificial General Intelligence (AGI) is the “sea of stars” in the hearts of all practitioners and believers, but it is too grand and distant. In the midst of incremental updates day after day, the industry needs a “story of the moment” that is more accessible and ignites instant excitement. “Vibe” came into being. It’s fresh enough, cool enough, and vague enough that anyone can project their anticipation into it. It promises a shortcut, a magic that bypasses boring training and goes straight to the heart of creation.
This also reveals the second secret of the popularity of “Vibe”: it is an “undefined” open concept and therefore has the potential to go viral. Any concept that can be strictly defined will set a cognitive threshold and thus limit its spread. On the contrary, “Vibe” lies in its ineffability. When you say, “This design is Vibe,” no one can argue with logic because it’s about subjective feelings. This ambiguity makes it unimpeded on social media and has also become a social currency for the identity of “insiders”.
But problems also arise. A concept that cannot be clearly defined cannot become a reliable cornerstone in serious landing scenarios. The connotation of this word is being diluted infinitely, and it seems that it is going to slide from a subtle metaphor to a slang word for the industry’s self-amusement.
Perhaps now we should go back to the scene where Andrej Karpathy originally proposed the concept. The most effective application of “Vibe” is not to let novices reach the sky in one step, but to make top masters feel like tigers.
People worship Rick Rubin as a totem, but often ignore his divine “Vibe”, which is rooted in deep intuition and aesthetic judgment tempered by decades of music-making experience. His “feeling” has a foundation. Similarly, Karpathy enjoys the exponential productivity of “Vibe Coding” because he is a top AI scientist in his own right. His “Vibe” – that is, his high-level intuition about code architecture and solutions – can accurately guide AI through tedious low-level execution.
“Vibe” is not a magic that everyone can get, it is more like an amplifier. It amplifies your existing professional depth and cognitive level. For experts in a field, AI becomes an extension of their productivity and experience capabilities, turning their intuition (Vibe) into reality at an unprecedented speed. For those who lack domain knowledge, the so-called “Vibe” operation is more like conducting a random experiment with uncontrollable results, which may be a surprise, or more likely to be “technical noise” and “creative garbage” that is difficult to maintain.
Vibe may be a good place to go when we talk about what it feels like to use AI, but when the term is used to describe products that require more accurate and reliable delivery, we need to take it more seriously.
