The best note-taking method for STEM students (it's not Cornell)

Every "how to take better notes" article on the internet recommends Cornell notes. It's not bad advice — for the wrong subjects.

The Cornell method was designed for college students taking fact-heavy lecture courses at Cornell in the 1950s. The format — a left margin for "cue questions," a wider right column for notes, a summary box at the bottom — is genuinely brilliant for history, intro psych, vocabulary-driven biology, and any class where the test asks "what's the definition of X" or "who said Y."

It's also wrong for derivations.

If you're taking calculus, organic chemistry, mechanical engineering, or quantum mechanics, you're not memorizing facts. You're tracking transformations. A line of math turns into the next line of math, which turns into the next, and the test asks you to reproduce the sequence under time pressure. Cornell's "cue question" margin doesn't help when your "cue" is "step 3 of a 14-step proof."

So what does work?

The two-track method

The honest answer: most STEM students do best with a hybrid that runs two tracks side-by-side.

Track 1 — the working notes — is just the math. Pen, paper or iPad, full-width. You copy what's on the board, with arrows showing how line N becomes line N+1. You write down what your professor verbally explains about the math (the part that won't be on the slides), but you don't waste time formatting it.

Track 2 — the annotation column — runs in the margin or on a facing page. This is where you write things like:

• "Why did we factor this here?"
• "This is the trick from problem set 3."
• "Prof said this is the part most students miss on the exam."

The annotation column is essentially Cornell's cue questions, but for steps in a sequence rather than vocabulary. Two weeks later, when you're studying, you cover the working notes and try to reproduce them using only the annotations. If you can — you understand the derivation. If you can't — you've identified exactly which step you don't get.

Why this beats outline notes

Outline notes (the bullet-point hierarchy most people fall back on) are fine for survey courses but they actively hurt STEM learning. Here's why.

STEM exams test your ability to execute a procedure. Outline notes flatten procedures into bullet points: "Step 1. Step 2. Step 3." That feels organized but it strips out the part that matters — the reason step 1 leads to step 2. When you read your outline notes a week later, you can read the steps. You can't reproduce them on a blank page.

The two-track method preserves the logical flow as the primary record (Track 1) and pulls the reasoning out separately (Track 2). Both are easier to study from than outline notes.

What about labs?

Labs are a different animal. You're following a procedure that someone else wrote, recording observations, and submitting a report later. Three things matter:

1. The exact reagent / instrument names your TA used. The lab manual will say "concentrated HCl" but your TA might pour from a bottle labeled "6M HCl" — write down what's actually in the bottle.
2. Procedure deviations. If the manual says "heat to 60°C" but yours actually heated to 67°C because the hot plate ran hot, that goes in your notes. Lab reports want the actual conditions, not the prescribed ones.
3. Anything the TA says verbally. "The expected yield is around 60% — if you got 95%, you did something wrong." That sentence is the difference between a confident lab report and a confused one.

For labs specifically, the best note format is the lab notebook column structure: time / procedure step / observation / value. It's basically a spreadsheet you fill in chronologically. Most chem and bio departments have a specific format they want you to use — follow it.

Where AI fits in

One question we get a lot: "should I just record lectures and let AI take notes for me?"

Short answer: no — but yes for review.

Recording-and-AI-summarizing is genuinely useful, but not as a replacement for taking your own notes. Two reasons. First, the act of taking notes is itself a learning event — you process the material in real time, decide what's important, and translate the professor's words into your own. Outsourcing that step costs you the encoding. Second, AI summaries are everything-in-equal-weight by default. They flatten "this is on the exam" and "this is a tangent about my dog" into the same density of prose.

What AI is great at: reviewing what you already wrote. Upload your hand notes after class, ask "what did I miss?", and the transcript catches the parts you didn't write down. Or ask "summarize this for me before the exam" the night before — that's when condensing 14 weeks of notes into one document is actually useful.

This is what we built ClassMinds for, and it's the only thing on this page where I'll plug it directly: the iOS app records lectures, transcribes them, and generates summaries that pull from your uploaded materials, not generic AI training data. Free during beta if you want to try it.

The summary, since this got long

If you're in a STEM major, here's the short version of what to do:

• Two-track notes — math/procedure on the main page, "why" annotations in a margin or facing page.
• Skip outline notes for derivation-heavy classes. Bullets flatten the logic that matters.
• Lab notes are different — a chronological columnar format works best.
• Don't stop taking notes just because you can record. The act of writing is the encoding.
• Use AI for review, not real-time capture. Best moment: night before an exam, ask "summarize what we covered this semester."

None of this is novel. It's just that "use Cornell notes" is the wrong default for STEM, and the actual right answer takes about thirty seconds to explain once you know what people are doing instead.