NOTES FROM THE LAB is a dispatch of ideas, a diary, a journal of progress or failure, and a peek behind the curtain into the creative process of being a photographer, filmmaker, multidisciplinary artist, parent, and husband. I’m Ramon, and I work with still and moving pictures.

The vibe has shifted. After excruciating years of breathless murmuring about AGI, the machine god, singularity, the total wipeout of the job market, there are new and more nuanced opinions on the limitations of the capabilities of LLMs in their current form.

The debates about existential risk, universal replacement and societal implications aren’t gone (as they shouldn’t be), but for better or worse, the tools are here to stay. Genuinely capable tools with real potential, and real problems.

Beneath the flood of generic content, the slop, the Mecha-Hitlers and Jesus Trump, there is the realisation that LLMs are the first tool that leave no mark. The output can sound like you, can look like you, and this is precisely what makes the advent of LLMs into the creative field so ambivalent (I’m only speaking of what I know, thus not even mentioning the implications in the coding sector, for example) LLMs are not harmful by nature, but most of the time invisible by design.

A tool you can’t see working is a tool you have to watch more carefully.

The flood of slop is here, and inside the slop there is only one really relevant question: Did someone actually think and feel this? Is there a way to distinguish what is actually born from a lifetime of experience or just pattern-matched from everything that came before?

There is no direct answer to this question, and probably no way to ever find out. But I think there is a responsibility we carry, as creatives, and it starts with us.

It starts with questioning our tools, to communicate our usage clearly and to forefront transparency in the way we are approaching things.

This is where a Code of Practice comes in, as a declaration for yourself and others of a stance of the fundamental question that you have asked yourself: Where does my judgement as a human end and where does the machine begin?

By day, I’m a photographer and filmmaker. I use LLMs almost every day in research, in the edit, for the tedious everyday administrative work that used to steal the precious hours between getting my child to daycare and picking him up again.

I’m genuinely, but cautiously curious about what LLMs can do for me and for my business. And this curiosity is growing, as well as my knowledge about the implications, and the world of LLMs will always remain a very sharp double-edged sword for me.

On the one hand, LLMs are really good at summarizing contracts, auditing finances, saving me a ton of money on tailored software tools that took a copious amount of time and money just 2 years ago.

But LLMs don’t create, they recombine. And that’s not a temporary flaw, it’s structural, and inherent in the system. Everything they produce is drawn from what already exists.

Human creativity is not method, its source. New and inspirational work comes from a body that has lived, from joy and failure, from the weight of real experience, from every scar and weakness. Patterns are not experience.

A 2025 Columbia Business School study found that people couldn’t reliably distinguish AI-generated art from human-made work — but when told the origin, they valued the human work 62% higher. And a CHI 2025 study interviewing established documentary photographers found that text-to-image generation fundamentally fails to depict lived experience. The photographers feared it would compromise the integrity of their narratives.

Using LLMs also means acknowledging what they cost. Every query burns energy and water. Data centers are proliferating against growing public resistance, and the long-term effects on how societies produce and access knowledge are still unfolding. I don’t think the answer is abstinence. I think the answer is not using the tool like it’s free, because it isn’t.

Share

I decided to go fully transparent on my usage of LLMs in my daily practice, and where I draw my line.

My work

My work is deeply rooted in curiosity about the human condition, and the relation to the world that is surrounding us. I will always promote true human empathy, connection, honesty and a conversation among like-minded individuals. \

But the economics of this profession are changing, and the pressure to adopt faster, cheaper, more automated workflows is real.

I understand the pressure because I live inside it. But I chose this medium for a reason, and it wasn’t efficiency in the first place. I chose photography because it puts me in a room with another person. Because the camera is a reason to get close, to stay longer, to pay attention. The connection I build with the people I photograph IS the work.

Refine, never invent

In photography, specifically in post-production, I use AI based tools to clean and refine what’s already in the image. They may never add what wasn’t there.

Photography has never been a neutral record of reality. Every frame is a choice: what to include, what to leave out, how to see. That has always been true since the invention of the medium, but there is a difference between the interpretation I bring to a moment I witnessed and created, and the insertion of something I didn’t. That difference is the line.

The idea is always mine

I don’t let LLMs generate creative concepts, shoots or projects. Projects emerge from experience, from encounter, from an intrinsic impulse to create something I’ve actually seen or felt. Ideas come from experimentation, and the relentless commitment to child-like curiosity. There are no shortcuts to ideation.

Resistance is the point

LLMs remove friction. That’s their selling point, and sometimes their value. But friction is necessary and essential for doing any creative work.

Reading a book means sitting with resistance, building connections, tolerating what you don’t understand. Reading the summary gives you information, not knowledge.

The same is true for creative thinking: the slow, uncomfortable process of finding your own answer is what makes the answer yours. Skip the resistance, and what comes out is weaker.

Cal Newport calls this “work slop”: output that feels efficient to produce but delivers less value to the people it’s meant for. And ultimately less value for yourself.

LLMs edit, never write

English is not my mother-tongue. I use LLMs to refine clarity, grammar, and flow. But everything that I write and that contains original thinking and ideas, from emails to treatments and essays, will always be written by myself. I use LLMs as a thinking partner and editor, but never as a ghostwriter.

My brief to them is simple: work to my code, not your own. You don’t introduce creative directions I haven’t defined. You ask questions, challenge assumptions, find contradictions.

But the impulse is always mine. Don’t invent, clarify.

Don’t replace, sharpen.

Studio Assistant

LLMs act as digital studio assistants. The administrative weight of running a creative freelance practice is real, and it competes directly with the time and energy that actual work requires. LLMs help me in select cases to automate tedious and repetitive tasks, specifically tasks that don’t need human oversight, and get a faster result with less time spent.

Footnote:
This code is not finished. It will change, as the technology evolves, as my practice changes and as I learn where I was wrong. I reserve the right to update it as I move along, and I will say so when I do.

A permanent version of this code lives on my website, publicly available for everyone to see. Feel free to grab it and adapt it for yourself.

See you next time.

xoxo,
Ramon

NEW WORK

New campaign work for Blacklane, with Anomaly and It Takes Time Berlin,
up now on my website: ramonhaindl.com/work/blacklane