While much of the world fears that artificial intelligence will eliminate thousands of jobs, there is a group of programmers who are pushing in just the opposite direction.
Not only do they accept AI writing code for them, they want it to do it forever. For them, it is an opportunity to redefine what it means to «be a programmer» today, not a direct threat to their job.
In recent years, the conversation about AI and employment has been filled with multiple warnings. Reports draw scenarios in which office tasks, administrative jobs, creative and technical positions can be partly replaced by increasingly capable models.
Programming has not been left out of that equation, with tools that generate code and agents that solve tasks fueling the feeling that even writing software is at risk.
For many, artificial intelligence is perceived as a substitute that erodes salaries, reduces teams, but above all, leaves humans in a residual role.
From writing code to giving instructions
In parallel to this concern, an opposite current has emerged within the development world itself. A group of programmers, many with high experience, do not simply tolerate automation, but accelerate it.
Its goal is not for AI to «help a little,» but to take care of the bulk of the coderepetitive patches, routine integration and maintenance of legacy projects.
From their point of view, writing code is the least interesting part of the job. What motivates them is deciding what problem is worth solving, how the product should work, what experience the user will have, and what limits must be respected.
It is important to mention that AI, in this scheme, is not a rival for developers, but rather the layer that absorbs what they consider mechanical and tedious work.
Andrej Karpathy y el «vibe coding»
This turn has a lot of weight where even Andrej Karpathy, one of the most influential voices in the AI ecosystem, published a message on X where he described a new way of programming that he called «vibe coding.»
In that tweet he was not limited to talking about advanced autocompletion. He proposed something more radical, such as giving the chatbot the responsibility of writing and modifying the code.
As well as accepting their changes without reviewing each line and focusing the human on describing the intention, correcting at a high level and adjusting the general direction of the project.
In this way, programming stopped being a job of detailed instructions and became a conversation with a system that generates the code underneath.
In a few months, vibe coding stopped being an occurrence on networks and became a topic of analysis, specialized pieces, as well as interviews with developers who are already working like this on a daily basis.
Why some programmers want AI to take their jobs
What these profiles want to get rid of is not the profession as such, but a part of the work they do on a daily basis.
For them, many classic programming tasks are perfect candidates for automation, such as repeating patterns, moving specifications into code, fixing bugs, moving blocks between files, adjusting syntax, or maintaining old systems.
By delegating that block to AI, they gain time and mental space to explore ideas, test different versions of a product, adjust features, and combine existing components in new ways.
Instead of spending weeks building a first version, they can iterate in several directions in a few days with agents that build, test, and fix the software almost in the background. From the outside it may seem that they are asking to be replaced, but they see it differently.
They want to stop being translators between requirements and lines of code and become responsible for deciding what is built, why it is built and with what limits.
While workers in many sectors see AI as a direct threat to their livelihood, a portion of programmers push it toward the core of their work and celebrate every advance that allows them to delegate more.
That this attitude comes from leading figures and teams that already produce real software indicates that we are not just dealing with a fad.
If the vibe coding approach becomes consolidated, the advantage will no longer be writing perfect code as much as knowing how to direct systems that write it.
That opens questions about how new developers will be trained, who will have access to these positions, and how value will be shared when AI does the groundwork.
In this «upside down world,» programmers who want AI to take their jobs could actually be designing the work that comes next.