Now publishing · Issue 20

Machine learning
in plain english.

One essay per issue. Honest about uncertainty, generous with examples, and free of ceremony.

Featured series · 5 parts

Stupid-simple ML

The easiest way to understand AI without being a total nerd. Five short essays, no equations, no jargon.

5 essays~11 min total

Start with Part 1

What AI actually is (and isn't)
Forget the sci-fi. AI is pattern-spotting, dressed in a confident voice.
2m
How a model 'learns' (the kid-with-flashcards version)
Training is just guessing, getting graded, and adjusting. A million billion times.
2m
Why training data is everything
A model is a mirror of what it was shown. Show it junk, get junk. There is no shortcut.
2m
Why LLMs Hallucinate
Why do LLMs hallucinate? Because nothing in the machine scores an answer for being true — only for sounding likely. The confident wrong answer is the system working as designed.
3m
How to actually use AI without losing your mind
A practical playbook. Use AI like a fast intern, not an oracle. Verify the load-bearing bits.
2m

Latest essay

★ Lead essay·September 16, 2026·5 min

Tokens and Embeddings

A token is a chunk of text; an embedding turns it into a point in space where nearness means similarity. Together they explain why models miscount letters, fumble arithmetic, and still find the right paragraph.

Read essay

llmbeginnerintuition

September 16, 2026·5 min

SGD vs Momentum vs Adam

All three follow the slope downhill; they differ only in what they remember. SGD remembers nothing, momentum remembers direction, Adam remembers direction and scale — and one of them has a bug you should know about.

Continue →

September 16, 2026·4 min

Overfitting vs Underfitting

Overfitting is memorising the training data; underfitting is never learning it. Both fail on new data, the fixes are exact opposites, and one ten-second check tells you which one you are looking at.