August To-Dos

One of the most influential courses I took during PAIR was Systematization. At first glance, this term appearing at an AI-focused camp might seem odd or even boring—that’s certainly what I thought. I had been familiar with the concept of systematization for a long time, which essentially means arranging things into an organized system. For example, setting up a daily routine—waking up at 9 am, drinking a cup of coffee, ...—counts as systematizing your day.

But what made this course special? Why dedicate an hour-long session to discussing this? After all, everyone knows how to set up a daily routine or organize other procedures, right? The answer became clear as the class unfolded.

For me, and I believe this applies to many others as well, the purpose of attending lectures isn’t just to learn facts or methods—after all, it can be inefficient to keep everyone on the same pace, and you can always learn by reading Wikipedia. The true value of attending classes lies in how you think and the discussions that arise with the teacher and classmates. Therefore, a great class should involve a lot of discussion.

This brings us to the key question: what was the real benefit of this course? It was simple yet profound—we sat there quietly and created our own systems to overcome bad habits. This turned out to be especially challenging, not just for me but for everyone, because it required deep thinking rather than passive listening. Not only did we need to construct these systems, but we also had to test them through experimentation. As I sat there, unsure of what to write down, memories of my previous attempts at systematization came flooding back.

People’s habits change frequently, and one of the most significant shifts is the urge to do something, to be productive and not waste time. I’ve often experienced this feeling—realizing how much time I’ve wasted and wanting to make a change. However, instead of actually opening VS Code and typing out some code, I’d start drafting my daily routines for the next week, hoping my future self would balance out my current regrets. But the worst part was when I never followed through on that routine, falling back into my old habits when the days arrived.

What have I been doing! Sitting in the classroom, reflecting on those days, I felt that familiar urge to do something again. Set up another productive daily routine? No! After all these attempts, I’ve learned one important thing: I don’t like to be restricted (how ironic), even by myself. After five minutes of focused thinking, I came up with an alternative approach to systematization: what if I scaled up the time frame to a month? At the beginning of each month, I could list the things I wish to achieve during that month. This static list would be based on my current interests, allowing for flexibility—I could do more or less depending on my motivation. However, doing less would reveal how many short-lived enthusiasms I have, which could still be a pitfall. Yet, this plan seemed more feasible because I could post these goals on a blog, each task coming in the form of a checkbox, and at the end of the month, I’d check off the ones I managed to complete.

With this plan in mind, what’s next? Experiment! Below is my first attempt at monthly systematization. I’m still glad that I pushed myself to try this out a week later (we had the course a week before). If you’re reading this blog after August 2024, you’ll also find the outcome below!

A list of things I hope to do in August:

• [ ] Finish reading "Our Mathematical Universe" by Tegmark.
• [ ] Learn more about information theory and KL divergence (to truly understand the basics behind).
• [ ] Write a blog on how to appreciate things.
• [ ] Learn more about theories of consciousness.
• [ ] Dig into the field of AI safety.
• [ ] Finish the methodology section for single cell perturbation paper.
• [ ] Start Drafting methods to improve the performance of RT-DETR on small object detections.
• [ ] Read through the full-length article on A Mathematical Framework for Transformer Circuits
• [ ] Finish reading "Axiomatic" by Greg Egan.
• [ ] Finish reading "Homo Deus" by Yuval Noah Harari.