Recently, I spent two days drafting a long, technical message, trying to explain some details of my work. I revised it repeatedly to make it clear without turning it into a wall of text. In the end, I even used AI to compress the message by about a quarter.
The response I received surprised me: some people react negatively to any trace of AI, seeing it as something that filters or changes the “authentic voice.” In that moment, I realized how subtle the line can be between productivity and the perception of authenticity. It wasn’t about laziness or lack of effort — I just wanted to communicate clearly, but the medium affected perception.
This experience led me to reflect on a fundamental principle in software engineering: abstraction is everything. Compilers, frameworks, and libraries exist to let us focus on what really matters: function, design, behavior. Every new tool has its skeptics. Some have laughed at the meme “blame the compiler” or rolled their eyes at new frameworks. Every layer of abstraction faces resistance, yet each allows engineers to produce more, better, and faster.
🔗LLMs as abstraction
LLMs fit into this history as another level of abstraction. They are not fundamentally different from a compiler or a framework: they offload much of the repetitive or verbose work, letting us focus our mental energy where it truly matters.
This does not mean outsourcing thought — it means compressing iteration cycles. Writing prompts, reviewing outputs, shaping results: these are all engineering tasks. Responsibility remains human, as it should.
🔗Authenticity and tools
It’s important to emphasize that LLMs do not cancel authenticity. Human voice is not erased because a tool is used. Authenticity is not the absence of tools; it is the presence of conscious decisions.
Every generated output requires judgment, oversight, and reflection. LLMs don’t think for us — they force us to think better, faster, and more precisely.
Of course, some people may react negatively to this idea. That’s understandable: historically, every new layer of abstraction has been perceived as a loss of control or a replacement of competence. But engineering is not about the medium; it’s about the outcome, the quality, and the responsibility. Embracing a new tool doesn’t mean giving up skill; it means using the right tool to achieve the best possible result.
Reflecting on my experience, I realized that the initial feedback wasn’t a personal rejection but a cultural reaction to the medium used.
There’s something paradoxical about rejecting LLMs on the grounds of “authenticity” while working in a field entirely built on layers of mediation.
- We write code through compilers.
- We communicate through protocols.
- We abstract hardware behind APIs.
Engineering has never been about raw, unfiltered expression. It has always been about controlled transformation.
If abstraction were the enemy of authenticity, modern software wouldn’t exist.
Tools don’t make engineers less human.
Refusing to understand tools might.
Abstraction has always been controversial. It always will be. But engineering is about leveraging the tools at our disposal to produce quality work. That hasn’t changed — only the tools have. Compilers, frameworks, languages, and now LLMs: the story is consistent. Tools evolve. Engineering remains.
🔗The real concern: ownership
There is, however, a legitimate concern worth separating from the noise.
The real question is not whether LLMs are “authentic.” The real question is: who owns the means of computation?
If the tools that increasingly shape engineering practice are controlled by a handful of corporations, if there are no viable open and competitive alternatives, then we are not talking about authenticity — we are talking about market structure and power concentration.
A free market requires real alternatives. An open ecosystem requires accessible infrastructure.
Criticizing engineers for using the most effective tools available misses the point. If we want a different landscape, we should build it.
The problem is not abstraction.
The problem, if any, is ownership.
Tools evolve. Engineering remains.
Markets, however, must remain open.