How to develop the most important skill for AI

A guide to building better judgement in AI by learning how to read papers.

If you want to work in AI or with AI, you’ll have to read papers. The space moves too fast, and by the time breakthroughs hit most blogs, YouTube channels, or company press releases, the space has already moved on. Even when people like me cover the cutting edge as it happens, you’re still seeing it through a filter. My judgment shapes what I emphasize, what I skip, and how I frame it. I might dismiss something as obvious that’s actually a common misunderstanding in the wider community. I might focus on speculative possibilities because they’re intellectually exciting, while skipping over the fact that they’re commercially dead on arrival (it’s why

Cameron R. Wolfe, Ph.D.

,

Sebastian Raschka, PhD

and I have different breakdowns on Llama 4, even when all of us agree on the basic facts).

Put another way, commentary is valuable, but it’s biased. The only way to get the raw signal and make sure you spot what is important to you in your context is to read the research yourself.

Unfortunately, this is very hard. Most people end up making one or more of the following mistakes when trying to read research papers:

• Drowning in every technical detail and every tiny derivation instead of understanding what’s important.

• Quitting halfway because the notation and math feel impenetrable.

• Relying blindly on YouTube summaries and blog posts without testing their own comprehension.

• A cousin of the above— using ChatGPT to summarize the paper, asking a few questions, and stopping there.

If you’ve ever finished a paper and realized you still didn’t know what it actually said, it’s not your fault. Papers aren’t written to teach you—they’re written to impress reviewers, make big claims, and they all assume context you don’t have. They can be extremely challenging to read for anyone who doesn’t already have an AI PhD.

That’s a story I can speak to personally. As I’ve talked about before, I am completely self-taught in AI. I had no academic background, no peer group to learn from, and no roadmap when I started. This article will share the system that I’ve spent years refining. I use it to read papers more effectively and identify the important takeaways, both to build systems currently and to anticipate what comes next. If you adapt it, it will help you build your own judgement around research, instead of relying on someone else’s.

If you make money from your judgment about AI, then you can’t avoid reading papers yourself. This guide will be a practical, no BS guide to how.

A preview of what’s coming

To access the full article—and all premium breakdowns going forward/written prior—upgrade to a premium subscription below.

Each piece is rigorously researched, built from firsthand signals, and written to make you sharper than the noise. It takes a long time for me to compile, analyze, and verify research. If you believe deep insight deserves support, become a premium subscriber to allow me to keep doing the same.

Flexible pricing available—pay what matches your budget here.

Supporting this work doesn’t have to come out of your pocket. If you read this as part of your professional development, you can use this email template to request reimbursement for your subscription.