Simpler Than It Seems: What DeepSeek Is Trying to Say
DeepSeek's harness has four words on the poster and a long paper behind it. The paper claims something much smaller, and much more interesting.
7 minute read
Categories
6 pages
Artificial Intelligence
Why DeepSeek's Harness Made Me Notice Every Wall in My AI Tools
AI made building your own tools nearly free — which quietly changed an old argument about what should be open.
11 minute read
Predictable Agent Trajectories: Agent Trajectories as Retrievable Context
How to stop AI agents from veering off course by giving them a memory of successful strategies - no retraining required
12 minute read
Is AI Making Us Dumber?
A GenAI engineer's take on the MIT study claiming heavy LLM users show reduced brain connectivity, and my personal take on the issue.
9 minute read
Exploring AGI and the integration of System 1 and System 2: JEPA, Transformers, and Dynamic Memory
An in-depth exploration inspired by Yann LeCun and Bill Dally's GTC 2025 discussion, detailing my thinking process on integrating JEPA with transformer models and developing dynamic learning and memory management in AGI.
4 minute read
Why JEPA Might Just Be the Missing 'System 2' in AI
A casual yet insightful exploration into JEPA, system 1 vs. system 2 thinking, and how AI might finally learn to stop hallucinating.
4 minute read