I am sending my BlueDot’s AI Safety: Governance application when I thought, perhaps, I can share my application here – it’s not perfect, but it’s reflective (even this application ‘essay’ was different than the previous one I sent for AI Safety: Alignment around 4 months ago – that shows some growth and I love to see that).
You can read my application to Intro to TAI (that I got accepted) too.
That (reflective despite imperfection) is a testament that I love as someone who works in AI Safety: that yes, LLM can/ could model after me (human), but I have the bigger agency to evolve, to be me, to learn from my pain, my failures, and sometimes, just be. Soaking in my being-ness.
With that intention, here is my application. Enjoy!
(TLDR: skip these all to Reflection section if you don’t care so much of the application essay haha)
How do you expect this course will help you contribute to AI governance?
You should specify:
- Concrete career steps you’ll take after the course
- How the course will help you take these steps
- How this will contribute to AI safety, and ideally AI governance
I’ve enrolled (and soon to finish) BlueDot’s AI Safety course where upon learning about compute governance in Week 7 (Technical Governance), it nudged me to not only researched upon it more but also made me applied to PauseAI’s compute governance volunteer researcher as well. My context would be to study how Malaysia (and SE Asia) is/ are positioned in compute governance, where upon more reading, made me realize the peripheral impact of these countries in compute governance – and motivated me to (1) research more (joining Dr Ana Valdivia’s talk on cloud, ‘interviewing’ Malaysians who are in semiconductor industry) and (2) thinking of establishing PauseAI Malaysia chapter (mainly for reading & study group).
Hence, I’ve contemplated to apply to BlueDot’s AI Governance course on two reasons: (1) ongoing live support (with the live discussion) while I’m undertaking the research, and (2) to solidify myself in this area: where as a small player in Malaysia now, I don’t have the luxury to just do technical AI safety but I need to know how to communicate about it & advocate (write, do videos) about my AI safety research. Fast-track AI Governance may not be as great as I’ve envisioned myself to engage deep in this, and as my AI Safety is now on Project phase (and AI Govt’s early weeks would be topics I’m already familiar with), I’m preparing myself for this goal.
These two prongs (research & advocacy) are the reason why I’m applying to this AI governance – directly for AI safety research I’m doing and directly impact AI governance efforts in my community. Ultimately and explicitly, I also want to use this opportunity (and this year) to solidify my application for Chevening/ international scholarship at the end of 2025 – as this field is highly needed for my country & community too.”
How have you engaged with the AI governance field so far?*
This could include things like:
- Events you’ve organised or attended
- Projects you’ve worked on
- Blog posts you’ve written
- Resources you’ve read or watched
“Projects: I’ve started doing entry works/ reading for the compute governance and from AI Safety’s Technical Governance session, I’ve started to draft possible topics to research & write more:
1. (e.g. Primer to Compute Governance for Countries like Malaysia (people think it’s not as important as we dont have Frontiers AI companies);
2. Compute Governance for non-Tech People (I’ve noticed that some Malaysians I talked to, could see AI Safety importance when I shared about the data centres to be built here in Malaysia + military case (eg Lavender) that’s being raised in BlueDot’s Intro to TAI)
Resources & events: I’ve started to read and talk to digital governance policymakers (from ISIS Malaysia, from Oxford Internet Institute), following up Dr Ana Valdivia, Prof Vili Lehdonvirta, Dr Luke Munn, and reading on related semiconductors books (The ASML way, and the Rare Metals War) as well as subscribing to SemiAnalysis newsletter.
I’m keen to extend my sharing last year about AI Safety framework in Islamic framework (for the consideration of Malaysia’s Prime Minister) this year but deeper.”
What skills have you developed that could be used to advance AI governance?*
This question allows us to place you in a cohort at your level of policymaking and technical expertise. You could tell us about:
- Programming projects or ML courses you’ve done (at university or self-taught)
- Government, think tank or similar work related to policymaking, especially on technology
- Politics, international relations or law courses that gave relevant skills
I’ve 3-year experiences working as a freelance rapporteur with Selangor’s (a state in Malaysia) policy think tank on care economy, child education, and digitalization for women-led SMEs. I’ve been contemplating & reached out to one policy friend about my intent in compute governance too, since then, to see the loophole and significance of this topic and the initial idea to extend my previous works before (given that I’ve networks to tap into).
I’m also one of Malaysia’s Parliamentary Fellow specifically on digitalization (back in Covid-19 time) and during the monsoon flood, I worked as tech lead for MUDA party (NGO side) in the volunteer HQ to help streamline processes and communication too. I also volunteer with other causes I love (eg Prof Normal Finkelstein for his eminent works on Palestines, and KuntumSayang for female refugee kids who didn’t go to school) where I love seeing my tech skills (though not at a bigger policy level yet) to be applied beyond the tools. I’ve helped consulted in the AI and humanities syllabus for a department in Universiti Teknologi Malaysia too in my spare time.
In term of ML courses, thanks to BlueDot’s AI Safety course, I’ve upskilled in replicating top AI Safety papers and even connected with several winners from AI Safety BlueDot projects to extend their studies with mine (eg Christoph Sträter, for his language-dependency LLM project that could heavily be used to extend the recent Alignment Faking study by Redwood, hence to be related to AI safety for Malaysian context (we have MaLLLaM, our own Malaysia’s localized LLM). I also took up Udemy’s ML class as I noticed I stuck a bit with technical concepts and tried my hands with ARENA.
TLDR: It honestly looks here-and-there (for this segment) but that is also one reason why I would like to enroll in this course: to further narrow down while converging my previous background into AI safety-governance. “
Tell us about one of your achievements you regard as most impressive, or that you’re most proud of.*
This should ideally demonstrate agency, ambition and an ability to get things done. If you’ve done something uniquely impressive in your work, or it is already highly relevant to AI governance we recommend you explain this.
If not, we recommend you explain your contribution to something outside your job. Strong candidates often describe projects that require unique skills and dedication, or projects that benefit the public or a large community.
“That I dont wait for the ideal time. I’ve recently pitched myself and ideas to National STEM Foundation (Malaysia) as I did talked to the vice president early last year about AI@kampung (literally, AI@village) and I’ve reached out to a friend who are in social enterprise to ask her to be cofounder for this AI governance-safety project.
I noticed that due to low resources in Malaysia – we may need to do AI Safety-Governance as one, at least for the introduction and at the start of the pipeline. I’ve been sharing BlueDot’s courses to my circle of friends, and I really intend to share about it more (but in my own way) this year and more publicly (with my Tiktok to my followers). I’m proud of how I am still doing this for my 3rd year from my own initiative and interest, and to use whatever ‘small’ achievements I’ve had (winning Top 3 1337 with AI-based apps idea, freelance rapporteuring, and others) for this goal. “
Reflection: throughout my career, a lot of people have been perplexed on “what you are actually doing, Shafira?” and through those questions, strangers and non-strangers, I’ve enjoyed every time I could answer:
“I’m being Shafira.”
To be honest, as I’m approaching 33 this year, I reflected a lot on how Allah pays me my teenager years in my 20s – to really explore & He sustained me all those years so sufficiently. If I could be vulnerable, throughout those 8 years of mine working in ‘wonder’ like this, only 2024 was the hardest financially (as 2 projects backed off and affected my finance for half of the year).
It’s not a testament (like “Hey you definitely can do it too”) so much (I could be having privileges others don’t), but I think at least my position is to be a friend if anyone wants to take some risk in wonderment.
To explore and pursue a question, while doing whatever we’re doing – that goes beyond the form.
We can be working full-time while still so dedicatedly embark your big personal question. For me, for many years, it has been about human.
Interestingly, that pursuit of mine didn’t bring me to psychology (I did thought about that, but I realized it didn’t suit my introversion as much), nor pure philosophy, nor other ways.
I picked AI safety. That overlaps between philosophy, technology, and in some ways, business/ startup.
That zeal of freedom & responsibility: the balance to pursue something while also remembering the responsibility we have.
And as I said to my partner, AI always tell me how lonely human could be. That’s what I see every time I see the computer, the coding, and the risks revolving it.
And he said, “That’s so beautiful.”
And that’s the deepest reason why AI Safety.
Leave a Reply