Reading, Writing, and Arithmetic using AI, While Staying Sovereign Using Obsidian
TLDR: Don’t be a Luddite. Be a Context Engineer.
Neuroplaticity starts to decline precipitously in our mid 20s. This window of time is when the average normie finds a job after school and starts debating if they want to commit to their situationship. At most they may read one Barnes & Noble tier pop-book a year and at best they are functionally illiterate.
Learning as a concept has been defined historically by standardized testing within the four walls of brick and mortar—soon to be purged from memory once graduated. Now learning is being offshored to Chat GPT. Just like other disruptive technology, Artificial Intelligence is here to stay.
Technological Convergence
Technological Convergence is the Dungeon Masters utopian ideal. With data as the new gold, and algorithms shaping us, any thinking person will need to find their thresholds within virtual space. We will also need to find our barometer when it comes to convenience within cyberspace.
After experimenting for sometime with dumb phones, Linux, and completely immersing myself within Apples walled garden I think i am finding solace within the shape rotation process.
So here is where i am with the blog moving forward and how i am integrating local models into my Obsidian vault.
I was inspired by Corey Doctorows method of object permanance and links on his blog https://pluralistic.net/. Ive also seen bloggers start with a TLDR, as a sort of abstract, which i will also integrate into my blog template moving forward.
Context Engineering Disagreeability into Obsidian
I wish Obsidian was open source but at least everything is in markdown and is easily accessible if i need to move to another platform. That said, I have been experimenting with Qwen 3.5B using an Obsidian extension known as Copilot. PDF OCR is only available using the paid subscription model. So i am looking for a means of converting PDFs to markdown and ingesting them that way. Things are moving quickly in this domain so I wouldn’t be surprised if another vibe coded competitor comes to market so i dont have to fork over $300 for a personal ask Hal9000-Jarvis-Jeeves-Clippy-Cortana-Jarvis AI to call up relevant concepts within my personal knowledge database.
Definition: Context engineering is the discipline of designing and managing the authoritative information AI agents need to make accurate, reliable decisions. It goes way beyond writing prompts; context engineering is a systems discipline: Architecting how data, instructions, tools, memory, and governance rules flow to an AI model at inference time.
Sycophancy is a known bug (feature) within the latest flagship models and it seems people are training disagreeability and pushback into models using DnD alignment. Here is the data for the Kappa Model: https://forum.level1techs.com/t/persona-kappa-training-a-20b-llm-on-desktop-gpus/246817
As time progresses and the androids receive their psychophantic, behavior-therapy instilled GPTs, we as Blade Runners will need to become our own context engineers to protect our own thinking ability. Lest the Dungeon Masters sequence our minds.
Links 🔗: https://en.wikipedia.org/wiki/Technological_convergence https://www.forbes.com/sites/chuckbrooks/2026/02/01/emerging-technology-convergence-will-shape-our-future/ https://www.weforum.org/stories/2026/01/why-technological-convergence-should-be-your-growth-strategy/ https://huggingface.co/eousphoros/kappa-20b-131k