AI Policy

I take the use of AI very seriously. I understand four separate problems concerning the use of AI:

  1. Epistemically, the possibility of cognitive offloading and hallucination propagation
  2. Practically, mounting of dependency on corporate products
  3. Socially and legally, (a) the illegal collection and “vacuuming up” of unpaid labor domestically (b) quasi-legal HRLF among underpaid workers in the global south and (c) the hollowing out of minority communities due to AI data centers
  4. Environmentally, the use of water by cooling centers
  5. Morally, the tainting of subjectivity and responsibility through (1.-3.)

I use AI models despite these problems. Against (4.), I do not believe that like can only produce like, as the classic Platonist formula goes. No, we can all admit that capitalism is a social poison. Yet, Marx was the first to admit that the bourgeoisie responsible for capitalism were revolutionaries: they singlehandedly put an end to the backwardness of all “feudal, patriarchal, idyllic relations.” No, the bourgeois revolutions were necessary, morally correct, and progress in the consciousness of human freedom. Yet, today, the results of this progress are a most disgusting miasma. Like does not produce like, a priori. AI’s production may be rotten, but its has no a priori claim to immorality.

Empirically, “moral taint,” is a bourgeois sham. There is no ethical consumption under capitalism—this idea is so often repeated as to be parody. Almost every commodity we use was built on the backs of quasi-slave labor. We use these commodities anyway. Taint is a nonstarter.

Nor is there anything unique about AI datacenters relative to the rest of the economy. According to independent 2024 from UC Berkeley, AI data centers used 66 billion liters of water in 2023 with an indirect footprint of 800 billion liters. Compared against the latest estimates from the US Department of the Interior, about 245 billion gallons of water were withdrawn in the US per in 2020. This is approximately 338,500 billion liters/year, orders of magnitude higher than US data centers. AI accounts similarly account for a meagre 15-20% of data center usage. According to almost all available recent data, there is nothing environmentally sui generis about AI data centers—they’re as rotten as anything else we use. The same goes for social media use, which depend on sweatshops like almost everything else we first-worlders consume. Again I repeat: there is nothing environmentally or socio-legally, and thus morally, unique about AI.

Except, of course, the vacuuming up of training data. Regardless of its still-pending legality, I find the exhibition of AI-generated material in any content production disgusting. AI-generations have no singular place in material one can stand behind.

As such, my policy on the mere use of AI is as follows:

  1. No material produced by an AI model will ever be exhibited, displayed, presented, or incorporated in any piece of content I produce.
    1. A media object (text, video, audio, etc.) counts as “AI produced” only if it was the direct result of an algorithm.
    2. Toolchains incorporating AI tools, such as transcription services and coding assistants, do not directly produce media objects. As such, they do not count as “AI produced.”
  2. All material signed with “misteramazing” branding is hand-selected and curated by myself.
    1. The output of all AI-assisted toolchains undergo my review.
  3. All social media posts produced by myself are either (a) hand-produced or (b) have their status as AI-generated clearly labeled by myself.
    1. No YouTube video or TikTok will use text or narration produced by an LLM, except as mentioned. (See: Use-mention distinction.)
  4. I use AI as a tool to augment my creative abilities. I am primarily an academic and a writer. I have had visions since I was a child. My visual and aural content is the manifestation of these visions. I have used computer algorithms to assist me in bringing these visions to life since I was a child: I began using motion tweening at the age of 11, digitized my handwriting at the age of 17, and have used generative video synthesis since the age of 18. LLMs are only a more computational complex algorithm. I use them as such.
    1. AI-assisted toolchains may control the production of images, video, and audio through programmatic manipulation of photo editors, video editors, and DAWs.
      1. As of 8/19/2026, AI-assisted toolchains begin from my Obsidian library of over 5,350 notes and by Zotero library of over 6,800 entries. I am the origin of every toolchain.
    2. AI-assisted toolchains may summarize pieces of my text on my website. I alone decide when they do so and clearly such summarization at the bottom of such pages. I am the end of every toolchain.
  5. AI is used in research purposes exclusively areas of my domain expertise: social, political, and moral theory and history.
    1. All AI research is validated by myself. If ever an AI-generated finding is incorporated into my material, its source is read by myself.
    2. My research process begins from my spontaneous intuitions and visions. They float before me as ghosts, and I snatch them as drops of the dewfall. I write with a whimsy and mystery that I enjoy - it is one of the few things that staves off my melancholy. I do not research, think, or write outside of my areas of expertise: I am not a mathematician, physicist, botanist, biologist, or chemist. I am not a historian, sociologist, or anthropologist. I am a theorist, and I use AI to augment knowledge about which I have accrued tacit familiarity through a Great Books Program, which books I own and re-read, my master’s degree in Political Theory, and my ongoing dissertation work at the intersection of propaganda and technics.