# Now <div class="pills-container"> <span class="pill">Last Updated: October 18, 2025</span> <span class="pill">Location: Metro Manila, PH</span> <span class="pill">Inspired by <a href="https://nownownow.com/about">Derek Sivers</a></span>. </div> ## What's keeping me busy I always seem to be doing a hundred things a once. I'm trying to avoid this [busyness trap](https://medium.com/swlh/avoiding-the-busyness-trap-15b0d77829d1) though, but I'm still learning how. - I spend 90% of my time doing AI safety-related work. - Leading AI policy reform research at AI4PH, which is supported by [The Asia Foundation](https://asiafoundation.org/) - Contributing to the [MIT AI Risk Repository](https://airisk.mit.edu/) as a research collaborator. - Helping out [Gavin Leech](https://www.gleech.org/about/) in reviewing the progress of AI safety agenda in 2025. - Designing alignment benchmark environments as a continuation of [this project](https://docs.google.com/document/d/15zlRwVakF_iYSKgeasfOgS8GdBoyuFG7b_fWkGNIpiU/edit?usp=sharing) from [AI Safety Camp](https://www.aisafety.camp/). - Helping out [AI-Plans](https://ai-plans.com/) run events and build their courses. - Curating a [database of AI safety papers](https://aisafetypapers.com) and [hosting a weekly reading group](https://paperclipminimizer.club/) to explore those papers in depth. Feel free to join the reading group by the way, though message me first if you do. - I spend 10% of my time doing other things like going to the gym or reading a book. I should really work on increasing this to at least 20%. This current allocation is too little and would lead me to burnout. ## What helped update my mental models 1. AI systems do not need to be very very smart in order to disrupt the world to an irreversible scale: [What Multipolar Failure Looks Like, and Robust Agent-Agnostic Processes (RAAPs) (Critch, 2021)](https://www.lesswrong.com/posts/LpM3EAakwYdS6aRKf/what-multipolar-failure-looks-like-and-robust-agent-agnostic) 2. Most of the time, it's better to have a general direction than a specific goal: [Explore More: A Bag of Tricks to Keep Your Life on the Rails (Tekofsky, 2024)](https://www.lesswrong.com/posts/uwmFSaDMprsFkpWet/explore-more-a-bag-of-tricks-to-keep-your-life-on-the-rails) 3. Success means being at the top 1% of what you do among others at your level, and you get there by never letting yourself work within silos: [Tips for Empirical Alignment Research (Perez, 2024)](https://www.alignmentforum.org/posts/dZFpEdKyb9Bf4xYn7/tips-for-empirical-alignment-research) 4. Anything I do, especially learning, should not be done passively: [Social Tinkering: Why Collaborative Curiosity Beats Vibe-Coding (Morris, 2025)](https://blog.cosmos-institute.org/p/social-tinkering-why-collaborative) 5. Taking time to process stuff isn't necessarily bad and could be a superpower if utilized correctly: [You're a Slow Thinker. Now What? (Hatta, 2025)](https://youtu.be/WsKi3DMqi2g) 6. Asking "what happens next?" and "is the world in equilibrium?" forces the failure modes out of a policy proposal most of the time: [Scenario Scrutiny for AI Policy (Turner and Kokotajlo, 2025)](https://blog.ai-futures.org/p/scenario-scrutiny-for-ai-policy) ## What I'm still learning I've always been a self-directed learner, but I'm learning that feedback loops and mentorship are force multipliers. If I want to grow, these are things I need to work on: - **Getting ideas to work quickly.** That means learning how to scope an idea in less than a day, and prototyping them within the next 24 hours. - A lot of my progress in programming and experiment design happened during the month I was doing [ARENA 6.0](https://www.arena.education/). I'm still quite slow compared to the [top 0.001%](https://lenz.wiki/wrriting/20250611), but I'm getting there. My pace is probably that I can finish the ARENA exercises on time when I'm alone, but somehow slower when I'm pair programming. - So far, I'm still able to write a thousand words in ~1 hour, but I'm a humanities major so...that's not really impressive. BUT I'm getting better and better in prototyping experiments quicker. - A big reason why I think my learning rate accelerated was because I learned to have a system that I can follow when building prototypes (or at least relying on existing frameworks rather than reinventing the wheel). The other big part of it is *being* in a high-achieving environment. When I worked at [LISA](https://www.safeai.org.uk/), I get bugged with "what are you working on?" and "how did you do X?" questions every lunch and dinner. That peer pressure somehow fixed my work ethic. Since I'm back home, I have to figure out how to recreate that environment for myself. - Generally though, I'm still a big comms person and still excel at comms-related work that is also useful in research (i.e., writing, building decks, generating visualizations, etc.). - I'm getting back to policy research after 2-ish years, and I'm so far still very slow at scoping. It's taking me 2 weeks right now, but I should shorten this to a week or less. - **Shortening my feedback loop.** It has been brought to my attention that my problem is not that I don't have skills, it's the fact that I don't share what I'm doing. Since no one knows what I'm doing, no one is also able to give me feedback. Well, I'm changing that. Feel free to give me feedback [via a call](https://calendly.com/ramennaut/1-1) or [via this feedback form](https://bit.ly/FeedbackForLenz). Please have a low bar for reaching out. I need thoughts from human beings aside from myself.