How AI changed the start of an Ultralearning project

Hey everyone, this is Allan. I’m recording a new video for my challenge of posting a video every day until I reach 100 videos.

Starting Ultra Learning

Yesterday I talked about the ultra‑learning project I’ve just started. Today I want to expand on the title: how AI changed the start of an ultra‑learning project. I’m still at the very beginning, so that’s the only part I can speak about right now. I worked all day, and it’s already almost 9 p.m., so I’m tired and haven’t had a chance to dive in properly yet.

The problem I want to discuss is not knowing how to start studying a new area or topic. If you’ve been a self‑learner for a while, you’ve probably experienced this a lot. If you’re used to school or college, where a roadmap is given, you’ll find it hard when you have to build your own path from scratch.

Meta Learning

Scott Young, the author of Ultralearning, outlines general ideas for structuring an ultra‑learning project. In a recent blog post he said that if he had to update the book today, the chapter on meta‑learning would be the one he’d rewrite, taking AI into consideration.

Meta‑learning is the work you do at the beginning of a project: researching how the subject is organized, identifying the main topics, the core books, the experts, and any existing curricula. In short, it’s figuring out what you need to study to acquire the desired ability.

AI’s Role

According to Scott, AI makes this part a lot easier and faster, but it also creates an excess of information. It’s now simpler to find the “ground truth” from genuine experts and the prevailing paradigms in a field. However, he warns against relying solely on AI responses. I agree, you should do deep research, locate the experts, and then use AI to help you find that content, not replace it.

So I plan to start my own ultra‑learning McKinsey project by doing the research phase first, then searching for experts. For now I’ll follow the free “Crafting Cases” course I mentioned yesterday.

Lever Analogy

AI responses are, by definition, mediocre, they’re the kind of knowledge anyone can access. If everyone can do the same thing, I need to put in extra effort to stand out. I think of AI as a lever: you need solid fundamentals to multiply effectively. If you start with a weak foundation, the leverage just amplifies the weakness. Building a strong base first lets AI amplify good work, not bad.

Practice and Effort

Even if I wanted to rely only on AI, the McKinsey project will still require a lot of practice. I’ll try to apply what I learn with my clients. To use AI effectively, you should conduct deep research, identify high‑quality material, and find recognized experts who produce learning content. Then, before you begin the actual learning process, you need to gather those resources.

AI won’t replace the effort needed to absorb content. You still have to read a lot, do exercises, and practice consistently.

Improving English

I realize I still have many skills to improve, English being one of them. I’m not someone who knows everything, and I expect to fail many times. Recording videos like this, pausing, re‑phrasing, and re‑recording shows that it’s possible to get better. It only gets easier with effort and repetition. If my English improves by the end of these 100 days, that will be a real accomplishment.