AI is moving forward quickly, making life easier for the one and harder for the others. “Vibe Coding” was established in 2025 as one of the terms in this context, thus lets take the “Vibe” term a little bit further.

A few years back, a colleague asked me about my opinion on using Augmented Reality solutions in maintenance processes. My response was complicated, just as many of mine are:
I suppose the main question is, why does the technician require the AR setup? If he is a skilled tech, knowing what he’s doing, and AR helps, i.e. projecting veins onto a patients arm , AR is an awesome tool. The same applies for use cases, in which somebody has no other choice. My favorite example is astronauts, i.e. when repairing toilets here and here . They’re in space, if they don’t do the job themselves, nobody will. On the other side, AR can be utilized to enable unskilled people to perform jobs, they otherwise would not be able to. And this is were the risk comes in. In my opinion, people should not perform tasks for which they were not properly educated, as eventually they will still have the responsibility and take blame. Additionally, if something goes wrong, well, there was nobody there to catch it. So, on average people or rather employees in a professional context, should not perform tasks with AR, which they could not perform without AR.
Luckily this also covers the astronaut scenario, as I guess a trained engineer or rocket scientist would also be able to reverse engineer a space-craft-toilet, just with a higher risk of shit hitting the ventilation system…

Vibe Coding

Seemingly coined by Andrej Karpathy in 2025 , Vibe Coding describes a software development practice during which somebody, either a developer, a skilled developer or a random person, utilizes AI or an LLM to write a program as in source code or script. The big risk being, as with my AR example before, unskilled people generate a program, solely based on functional requirements. So while a skilled developer not only makes decisions on what a program does but also how it does it, this step can be completely missing in Vibe Coding. To be fair, this does not have to be a bad thing. If what comes out solves the problem, perfect! In return there is less in the way of things going wrong.
This is also not a new problem! Whenever somebody starts training a Junior, the Junior will make mistakes, the Senior will catch them. This especially applies to logical flaws, which can be pretty hard to find. If the Senior makes mistakes, we hope its not a logic flaw, and then try to catch it with static and dynamic code analysis. Eventually we hope to have limited bugs as far as possible. Nobody is perfect, so we accept certain gaps.
If vibe coding is performed by a Junior and then run through static and dynamic code analysis, we simply lack the human factor catching logic bugs and things the Senior has learned to be bad ideas.

Vibe Supporting

I would like to establish the term “Vibe Supporting” as an extension of already existing low budget support strategies: Support staff is not properly trained, but equipped with access to an AI System / LLM, instead of the infamous “knowledge base”. Talking to the customer, they type their understanding of the issue into a chat interface, and respond with what the LLM replies.
Just as with Vibe Coding and the VR example, human quality assurance is not applied, if the supporter has too little knowledge about the topic.

Human Washing

This brings us to the next term “Human Washing”. By actually having a human in the loop, the customer is not able to distinguish between working of somebody’s knowledge, or basically just having a fleshy interface with a computer. The supporter compensates for all quirks, like grammatical errors, incorrect words, wrong use of tense. Because, while the supporter might not understand the content, he definitely understands the applicable language.

Vibe Operations

The term “Vibe Operations” is the logical result of current AI system usage. Utilizing AI systems / LLMs as a natural language interface for systems during maintenance and daily operations, the human is not the one executing commands anymore. Only describing what he’d like, the AI systems interprets this and executes what it expects to be correct. The implications of this, well, a simply repetition of everything said before.

Is This Criticism?

No, its a warning, and the attempt to establish new terms for future discussions. Just look back at the use of “Cyber”, the resulting drinking games, extra boxes in Buzzword Bingo. “Vibe-*” is going just that direction.
From a technical perspective, many companies have learned, that shifting first level support to the cheapest country available, does not increase customer happiness. Many admins have learned, that using shi**y, non-deterministic scripts can result in major issues. Some people have already learned, that even the best current AI system can and will break major systems. Turning something into a Buzzword still allows companies to use it for advertisement, but at the same time creates a certain ironical touch to many things, making space for criticism.

So let’s go “Vibe-*”-Drinking.