# Next <div class="pills-container"> <span class="pill">Last Updated: October 18, 2025</span> </div> ## The next five years When people ask me what the next 3-5 years look like for me, I reckon they want an answer like "In 5 years, I want to be a lawyer" or "In 5 years, I want to be in a more managerial role." I feel uneasy answering that. I don't quite believe in job titles, but rather I look into the gravity of the contribution. In 5 years, I want to be doing the most impactful work I *can*. That would mean working on projects or products that affects people's lives at scale (and ideally for the better). Right now, if I stay in my home country, [the obvious thing](https://www.lesswrong.com/posts/Zpqhds4dmLaBwTcnp/trying-the-obvious-thing) would be to work in policy reforms.[^1] If I end up moving to an AI hub, then the next obvious thing transitioning to multi-agent safety research like sandboxing, evals, or doing [more work in technical AI governance](https://cdn.governance.ai/Open_Problems_in_Technical_AI_Governance.pdf). Now, these high-impact domains are also high-trust. To even be taken seriously among these circles, I need to churn out a lot of quality work and maybe even get a graduate degree.[^2] So perhaps, my answer is: **I want to spend the next 2-3 years building the career capital needed to help design better guardrails for emerging technologies like AI, then the rest of the next five years to position myself towards being the best in my own niche.**[^3] That might mean pursuing a graduate degree or an accelerated career ladder to eventually step into leadership roles where I can shape decisions that matter. If I want downstream impact, I need upstream positioning. That means building both credibility and capability so I can eventually shape how tools and policies are designed, and be able to contribute competently in my chosen field. ## Questions worth answering This space is reserved for questions and ideas I want to explore (or want explored by other people). This is a living document. The things written here reflects what I'm currently thinking about and how I might be thinking about them. I could be wrong about certain assumptions, as [[Ethos#Mistakes I've made|I've been wrong countless of times before]]. Note that I have not done a comprehensive literature for some of these so it would be good to check the existing literature before starting work on any of them or believing it as is. If you are interested in working on any of these ideas, notify me at `mail[at]lenz[dot]wiki`. ### Science of agentic evaluations I'm particularly interested looking into the [ecological validity](https://en.wikipedia.org/wiki/Ecological_validity) and [construct validity](https://en.wikipedia.org/wiki/Construct_validity) of eval setups to make them more reproducible. - **Developing multi-agent and multi-objective evals for both capabilities and alignment.** I did some [initial work](https://docs.google.com/presentation/d/1ePaTc4qq4Ec8eZQV-V4Ev1NfK5x-Ky3P8JmpwA2XDp0/edit?usp=sharing) on this in [AI Safety Camp 10](https://www.aisafety.camp/). I think this line of work is generally underdeveloped compared to single-agent benchmarks. We don't know much about how multi-agent or multi-objective systems align, but we also don't know much how they even develop certain capabilities since behaviors become emergent in these settings. - **Building interp-based and environment-based evals.** Compared to static evals, these are much harder to game (I think).[^4] There's already a decent number of environment-based evals that are being built, but interp-based evals (like using linear probes or crosscoders to do eval on a behavior) may actually be promising. [AbliterationBench](https://github.com/gpiat/AIAE-AbliterationBench/), which uses steering vectors, is an example of this. - **Debugging the evals ported to [Inspect](https://inspect.aisi.org.uk/).** I worked on this along with [Hugo Save](https://github.com/HugoSave) for our capstone project in [ARENA 6.0](https://arena.education/). In our case, we focused on Google DeepMind's Dangerous Capabilities Capture-the-Flag suite and found some bugs; and this was the most used eval (a.k.a. most maintained) at the time we were working on it. I would guess that there's probably a bunch more (and bigger) bugs in the 90+ evals that are ported to Inspect, especially those which aren't used much. - **Docent/Transluce for interp-/environment-based evals.** Making transcript analysis easier to do makes debugging evals easier to do. At the moment, current tools just don't apply to interp-/environment-based evals (at least those that does not rely on LLM transcripts). ### Field-building in Asia While frontier AI safety within western AIS hubs are obviously very relevant, I'm also quite concerned about AI safety field-building in Asia. The region is very diverse, and so policy interventions would look a lot different in this space, compared to what we see in the US, UK, or Europe. There's also way more compute deserts and compute south countries in the region, which can be a concern. My thinking is that whatever this region needs can be unlocked by policy reforms; but policy reforms are rallied by advocacy and field-building work. - **Direct work in policy.** There are many emerging opportunities to work in AI policy within Asian countries right now. I'm most interested in efforts within China, India, Japan, Singapore, South Korea, and Taiwan. Although the rest of Southeast Asia, especially Malaysia, Thailand, and Vietnam also seems to be catching up, so it's worth looking out for efforts in these spaces. I work in Philippine AI policy reform, and I chose to do this because there's relatively a lot of AI safety folks coming from the Philippines, and the country can be instrumental in field-building. - **Project-based upskilling.** Too many people seem to get stuck in theory and discussion groups. Environments where people can ship and get feedback is critical. In AI safety, we're working with way less manpower and funding compared to the rest of AI development investments globally. Maximizing this resource could be one of our best chances at ensuring responsible AI training and deployment. This is truer and more urgent for Asian countries, since most AI safety training courses are based in the US, UK, and Europe. - **Designing quicker feedback loops.** The gap between ideas and prototyping is getting narrower over time. But this sort of opportunity exists mostly for senior-level researchers and the small pool of junior-level researchers that they can take in. Young folks from halfway across the world usually don't have access to these types of opportunities. But infra that is designed for this like the [Apart Sprints](https://apartresearch.com/sprints) are promising. I think we should have more of these across different sub-areas of AI safety. Not everyone needs to be a full-time AI safety researcher. I think strong middle layers can help fill the vacuum between interest and expertise, and can actually move high-agency people towards doing research that is 4x more impactful that the normative output with more FTEs. - **Building a tech consultancy for AI safety.** I reckon that once policies are implemented, there would be more demand for AI governance experts and auditors. With that in mind, I think it's only logical to have some kind of McKinsey for AI governance. ### Non-AI safety ideas that I think are cool - **[Shoutout.io for feature requests.](https://shoutout.io/)** Okay, this may exist already. But wouldn't it be really cool to have a shoutout dashboard for feedback and requests instead of testimonials? - **Mobile-based AI agent builder.** Imagine [n8n](https://n8n.io/) but you can access it with your phone. I genuinely think there's a lot of tools that can be made for micro-entrepreneurs. Most of the time, software is too expensive or too complicated for their needs. I think there's a sweet spot that could target a bunch of people in this pipelines who don't have budget to sustain hiring manpower but would love to be able to scale independently as well. - **Vibe-modified UI libraries.** Imagine if you can basically modify [Magic UI](https://magicui.design/) like how you choose colors with [Coolors](https://coolors.co/). Then you can just download and paste them. All you have to think about is composition. That's like drag and drop web builders + vibe-coding tools, which genuinely sits in the middle of technical and non-technical building. - **Docent for policy research.** I don't know about other governments but the Philippine government churns out hundreds of bills every week. As a policy researcher, I would like to not manually read every one of those just to find loopholes or discrepancies. [^1]: Why is AI policy the obvious thing? It's really not given my skill set, but since I live in the Philippines and I still want to work in AI safety, I could argue that it is the best way for me to make impact in my field. There's not a lot of technical stuff going on locally, and the bottlenecks for those can be unlocked by (surprise, surprise) policy reforms. So if we want more local safety work, then we need reforms to make it happen. [^2]: Although I could argue against this point and say that given shorter AI x-risk timelines, graduate degrees are useless. I can figure out how to be the best of my field if I consistently produce high-quality work. [^3]: I initially thought I'd want to spend the next 5 years to upskill, but that's quite a long time to be upskilling for such an urgent issue; so I'll have to make pursuing an accelerated career path a feasible option. [^4]: This is so far my working hypothesis. I can be proven wrong.