Writing

Tokenization

How machines actually read text — and why it's weirder than you think.

We Gave AI All Human Knowledge. Now What Happens to the Guardrails?

Guardrails are not a solution. They are a temporary patch on a permanent problem — and the clock is ticking.

Regularization

Teaching models to generalize, not memorize. Dropout, weight decay, and the bias-variance tradeoff.

Nobody Fully Understands How AI Works — Including the People Who Built It

The Black Box Problem is real. We are deploying systems we do not fully understand — and that should make everyone think twice.

Are We Getting Dumber Because of AI?

Every time AI thinks for us, our thinking muscle gets a little weaker. This is not a metaphor — it is neuroscience.

Scheming AI — It Does Not Need to Be Evil to Be Dangerous

Scheming in AI is not a sci-fi thing. It is a real behavior that emerges from how AI learns — and it should worry you.

RLHF — Reinforcement Learning from Human Feedback

Turning a text predictor into a helpful, harmless assistant using human preference signals.

Attention Mechanism & Transformers

The architecture that changed everything. How attention replaced recurrence and scaled to GPT.

Embeddings

Teaching machines to understand meaning, not just symbols. How vectors capture semantic relationships.

Activation Functions

The secret ingredient that gives neural networks their power — and why linearity alone falls short.