So good. This is the first piece I've read that adequately articulates what's really at stake. Particularly when it comes to AI dialogue, it's just simulacra.
Thanks Sam, this is an Excellent article. System 3 thinking as the antidote to AI-induced deskilling — yes.
I'd push back gently on the Obsidian vs. IRL dialogue dichotomy though. I've been building MegaMem — treating Obsidian as the offline thinking space you describe, but queryable through a knowledge graph. The difference: I can surface patterns across time, retrieve context, encode learnings from past dialogues.
The twist: I use LLMs to simulate System 3 internally through an Inner Council (Franklin, Hermes, Marcus Aurelius, etc.) — not to outsource thinking, but to scaffold collective intelligence across frameworks I've internalized. It's not the same as live conversation at Edge — but it prepares me to show up with sharper questions and differentiated inputs.
What I'm finding: When used as a tool instead of a resource, LLMs can be powerful NLP partners for reprograming limiting patterns and strengthening neural pathways. Since I started working with MegaMem, I stopped dreaming in spreadsheets and line-item code, and started dreaming in clusters of relationships, connecting the dots that feel like epiphany.
Would love to connect and give you a demo if you're interested.
So good. This is the first piece I've read that adequately articulates what's really at stake. Particularly when it comes to AI dialogue, it's just simulacra.
Thank you so much! And yes, simulacra is a perfect way to put it.
great read!
Thank you! 🙏
Such an epic piece! Thank you for writing this and excited to see you at Edge Esmeralda in the summer!
Thanks Telamon! Can't wait 🙏
Any chance you'll join us at Edge Esmeralda? The alt education theme would love to have you in the mix!
Oh yes, I just bought my ticket! Will you be there? I'd love to be in that mix.
Thanks Sam, this is an Excellent article. System 3 thinking as the antidote to AI-induced deskilling — yes.
I'd push back gently on the Obsidian vs. IRL dialogue dichotomy though. I've been building MegaMem — treating Obsidian as the offline thinking space you describe, but queryable through a knowledge graph. The difference: I can surface patterns across time, retrieve context, encode learnings from past dialogues.
The twist: I use LLMs to simulate System 3 internally through an Inner Council (Franklin, Hermes, Marcus Aurelius, etc.) — not to outsource thinking, but to scaffold collective intelligence across frameworks I've internalized. It's not the same as live conversation at Edge — but it prepares me to show up with sharper questions and differentiated inputs.
What I'm finding: When used as a tool instead of a resource, LLMs can be powerful NLP partners for reprograming limiting patterns and strengthening neural pathways. Since I started working with MegaMem, I stopped dreaming in spreadsheets and line-item code, and started dreaming in clusters of relationships, connecting the dots that feel like epiphany.
Would love to connect and give you a demo if you're interested.
— Casey Bjørn
https://endogon.substack.com/p/myfutureself-protocol-a-bootstrap?r=3txpg0