Last updated: May 2nd, 2025 AD in Somerville, MA by Nic Weyand!

NicWeyand.com proudly does not use AI in any writing. All posts on this site are 100% my work.

However, I will occasionally use AI as a sounding board for my works sometimes. This is the extent of my use, and I do not copy and paste anything the AI gives me into my works. This whole site is written by myself, Nic Weyand.

To be clear on any nuances, here's what "Minimal AI" means:

• I use spell checkers, grammar checkers, formatting tools inside of programs like Scrivener, Ghost CMS, LibreOffice etc.
  • This is technically AI, and for the sake of transparency, I will state their use, even if it feels obvious.
• I do not prompt ChatGPT, Claude, DeepSeek, Llama, etc. to write anything for me. All the writing & research is my own. I have not, and never will, ask any AI model to write my work. 100% of everything on this site (nicweyand.com) is written entirely by myself, Nic Weyand.
• Occasionally I will feed drafts into an Ollama Deepseek R1 model I have running offline to give feedback and opinions. Essentially I use offline models to act as sounding boards.
  • I am in no way obligated to use anything they say, its just nice to use as a tool sometimes.
  • I have moral and ethical qualms, given these models are all built on plagiarism. My only defense to using them is this: they exist, and they are useful. This is a bad defense, I admit. I am still going to occasionally use offline models.
    • I am uneasy with using them given the plagiarism issue, and frankly they aren't very good as-is. Using offline models at least mitigates environmental damage, versus ChatGPT and large, online-only models that are wildly inefficient.
    • If you were not aware: ChatGPT in particular has been seemingly designed to require an insane amount of electricity to work, and massive water reserves to cool data centers.

The computational power required to train generative AI models that often have billions of parameters, such as OpenAI’s GPT-4, can demand a staggering amount of electricity, which leads to increased carbon dioxide emissions and pressures on the electric grid.

Furthermore, deploying these models in real-world applications, enabling millions to use generative AI in their daily lives, and then fine-tuning the models to improve their performance draws large amounts of energy long after a model has been developed.

Beyond electricity demands, a great deal of water is needed to cool the hardware used for training, deploying, and fine-tuning generative AI models, which can strain municipal water supplies and disrupt local ecosystems. The increasing number of generative AI applications has also spurred demand for high-performance computing hardware, adding indirect environmental impacts from its manufacture and transport. (Adam Zewe | MIT News | January 17, 2025)

• By opting for offline-only models, I hope to offset the environmental impact slightly, though I fully acknowledge training takes up a lot of resources, and it was built and trained off of stealing literally everything on the internet.
• Occasionally when I hit a wall, I can use them for help and they sometimes bring up something I hadn't considered before. That is the sole benefit, in my use.
  • I have no problem deleting Ollama and the models in question from my computer if this feels like a violation to my users. Given my current feelings of uneasiness, I might do that on my own altogether.
  • It is my policy to never copy and paste anything an AI gives me. I would consider doing so "AI writing for me", instead of me doing my own work. If the AI raises a point or counterargument I hadn't considered, I write everything responding to it on my own, in my own voice, and never write what it gave me directly.
  • I only use offline models (Currently DeepSeek, IBM Granite & Meta's Llama 4) via Ollama, as I find using online services like ChatGPT environmentally destructive.
    • Moreover, I do not wish to support OpenAI or Anthropic, as I find them unethical.
      • Anthropic at least tries to have some ethics built into Claude, but dang are they corrupted morals these days.
  • I have had ChatGPT and Claude accounts, and have used these services in the past. I have not, and will never publish anything generated by these platforms here or on any other platform bearing my name, as I see their use mostly as experimentation and getting familiar with what's out there.
    • I find Claude to be better at generating code, but it is still AI, and the code is fairly worthless to my projects. It simply doesn't know what I want, or how to build it properly. The lack of full context, even in project knowledge, makes building projects properly impossible.
      • Just because it ran in your IDE does not mean it is good code.
    • I like programming! And I would rather do it myself, and know every bit myself.
    • I think "vibe coding" and similar concepts are antithetical to the artistry possible with programming, as well as the obvious technical skills. I believe this will introduce serious security holes by untrained people merely shipping code "because it works".

Okay, I think that basically sums everything up. TLDR for the illiterate!

TLDR:

• I use offline LLMs only, via Ollama.
  • Deepseek, IBM Granite & Llama.
  • I do not use ChatGPT or Claude, or similar online-only models.
• I do not use LLMs often. They aren't very good.
• When I do use them, sometimes they will read my work and offer suggestions, counterarguments and other things I hadn't considered.
  • I am under zero obligation to use anything an LLM gives me for my work.
• I do not copy and paste anything an LLM gives me, as I consider that cheating. I write all my own stuff, and do not let anything else write for me, or in my voice.
• I have literally zero problems deleting these models if people find they are uneasy with me using them, or have ethical qualms.
  • I do not like how the models are trained off stolen data.
  • I do not like the environmental impact of training off stolen data, and the excessive Co2 emissions and water used with every ChatGPT prompt.
  • I do not like Silicon Valley, generally, and don't particularly like supporting anything they do these days, so to the best of my ability I try and avoid them. Given their reach, that's difficult, but I do alright!