NoAI - A statement against generative AI

I am strongly opposed to any¹ form of AI being used, especially generative AI, be it text, image, video, or anything else. This page will serve as a living document and writeup of the reasons I am opposed to the use of generative AI.

Note: This document is not finished. Any grayed-out sections are yet to be started/finished.

Why?

For me there are more than a few reasons, but I’ll separate them into four major categories of concern: The human factor, environmental, ethics and societal, and financial. While the human factor and ethics are closely related, they feel distinct enough for me to separate them into two separate categories.

For simplicity’s sake, whenever I refer to plain “AI”, think Generative AI (LLMs, image generators, etc.). Machine Learning (ML) is distinctively different, and not comparable to GenAI.

The Human Factor

To me, the human factor is the most important argument against AI. It has the very noticeable impact of hampering your creative abilities, as well as affecting your critical thinking skills. It also creates a very false narrative that creating something is easy and requires no human input other than a prompt.

Content generated by AI also has a very distinct lack of personality, and follows very set typing patterns.

There is also the big issue of the “dead internet theory”. Chatbots are being used more and more instead of people visiting sites organically. This means content that someone spends hours, days, weeks, or even longer on isn’t actually being read at all. Further, a lot of pages block AI bots, making their content inaccessible to everyone using AI as their search engines.

Creativity

I used to be an avid user of ChatGPT back in 2022-2024. I used it for everything from generating quick bash scripts to writing professional emails. When I learned how bad AI really was (for all the reasons listed on this page), I stopped using it entirely. Since then I have noticed how much my AI use have impeded my own learning and creative abilities, and how difficult it has made things that were previously easy.

I will admit that this is very subjective and not very good evidence, but from talking to my peers (again, subjective) I have found an overall decrease in “thinking capacity” compared to before AI exploded everywhere.

Creates a false narrative that creating something is easy, automated and doesn't require human input other than a prompt (i can just create this with AI in 10 minutes).

Organic website traffic, "dead internet theory", sites blocking chatbots, hallucinations. Reading algoslop instead of things someone spent hours on. Lack of human creativity and yank, aka loss of authetic human slop

ai psychosis and similar adverse health effects

Ethics and societal

On the more ethical problems, and problems affecting all of society there is the issue of bots changing how we write, and schools having to use their time and energy on “beating AI” instead of doing teaching and improving teaching materials.

A good example of AI changing how we type is the em-dash. Before the era of chatbots, the em-dash was a sign of someone who knew how to type properly. Now, a lot of people are using it as an instant “aha!” when trying to detect AI content, even if said content was written completely organically by a human being.

It's also doing long-term damage to education because it's forcing schools to use their time and energy on "beating AI" instead of actually teaching.

one user vs whole society - "throwing trash in the nature" argument

Stolen content

Trained on stolen content

Emdash becoming a "forbidden character"

the further destruction of the web as a human centred social community

The erosion of trust in the visual medium

Environmental

Water use, power use, temp changes? Increasing power prices in northern norway, nrk article https://www.nrk.no/nordland/open-ai-og-aker-bygger-datasenter-i-narvik.-det-kan-gjore-stromprisene-fem-ganger-sa-hoye-1.17516127

Financial

Something something about the bubble, circular funding, hardware pricing, etc

Other reasons

Code maintenance, spam AI PRs

The erosion of documentation for coding is another factor that ive heard a bit

about Hallucination of sources in research, high AI use - https://doi.org/10.48550/arXiv.2502.00632 https://arxiv.org/abs/2403.13812 https://arxiv.org/html/2601.18724v1 https://www.youtube.com/watch?v=SEwiOykoXXc

Commonly used reasons for AI use

"I only use it as a search engine"

Improve your searching instead. Google-fu is a lost art, but a very important one.

"I don't know how to draw! Image generation lets me be an artist!"

No. Anything made through GenAI is not your work, regardless of how long you spent on the prompt to generate it. AI “art” is trained on stolen content (see #Stolen content). Nobody knows how to draw amazing art immediately, artists spend years and years on practice to hone their skill. If you want a piece of good artwork, pay a real human and commission it instead. Contribute to the economy, help others.

FAQ / Counterpoints

”But x, y, or z!”

This page isn’t meant to list and discuss every type of AI or reason against AI use. I have listed the biggest arguments against AI I could think of, however there are certainly fields which can benefit from the technologies and ML evolutions happening because of all this. Some of those are listed below, but again: This page is not exhaustive.

How about the medical sector?

There are plenty of researchers trying to achieve good things through AI and Machine Learning, especially within the medical sector there are some fascinating use cases:

AI and other previously inaccessible automated analysis methods have been used for detecting and diagnosing forms of cancer and other tumors. These are clearly groundbreaking pieces of research, however despite these I am still critical to GenAI use in medicine. This applies to text and chat services especially, largely due to the potential for hallucinations and otherwise inncorrect or directly harmful information, as well as the massive privacy implications in regards to HIPAA, GDPR, and other privacy frameworks.

These concerns obiously apply to more than just medicine, but that is where I am the most concerned due to how bad companies are at data security already.

“Any” here generally means generative — there are some more valid uses for AI such as medical purposes (see “FAQ / Counterpoints”), but the jury is still out on those usecases as well. ↩