My experience as a CLR grantee and visiting researcher at CSER

By Jaime Sevilla @ 2020-04-29T19:03 (+99)

Last year I applied and was offered a grant from the Center for Long-Term Risk (CLR) fund to conduct independent research.

Me and the Center for the Study of Existential Risk (CSER) of the University of Cambridge agreed that I would spend my grant period as a visiting researcher at their institution.

I am quite grateful to both organizations for supporting my work over the last six months. Now that the period has ended, I have decided to write a blogpost reflecting on my experience and achievements these last six months.

Experience

First and foremost, my experience applying and receiving the grant from CLR was as good as one could hope for. There was no major bureaucratic load, and CLR’s staff was very supportive, flexible and professional through the process.

I kept them updated monthly on the activities I engaged with, not as a requirement for the grant but more as a form of self accountability, keeping in touch with each other’s research and providing them with some useful information which I hoped would help them with future grantmaking.

As I started the visitorship at CSER I found a really welcoming group that went out of their way to make me feel included. As I arrived I was offered to share my current work and participate in their weekly meetings, and I always had people who would listen to my problems and concerns.

In CSER I found people whose reasoning I’ve come to admire greatly, and I’ve come off with a greater understanding of the world. I want to highlight Seán ó Héigeartaigh, Jess Whittlestone and Shahar Avin who are excellent researchers with very good epistemics and very goal driven.

As negatives, it was quite hard to move to Cambridge. The friends I knew from before who live here were quite busy and at times I felt very lonely. I think the short hours of light in winter contributed to me feeling down.

Some things that helped mitigate this were roleplaying with the friends I made at CSER, informal light therapy with a SAD lamp, joining the Cambridge MTG community, biking every day for at least an hour. Special props to Haydn Belfield for keeping me sane through these six months by being a great supervisor and friend, and to Sabin Roman for being a great friend and coworker.

The COVID19 pandemic caused me a lot of anxiety and stress, but overall my situation was much better than others’. I could keep working from home, I had stockpiled a lot of Soylent beforehand and my family is all healthy and safe.

Achievements

In the first week of my visitorship I discussed with my supervisor my goals for the six months. This were the five things I set for myself as goals:

• Write at least two academic writings: a report on moral patience and another one on quantum computing
• Figure out what to do after my visit ended, and particularly apply to some PhD programs
• Learn about CSER’s research and collaborate with people here
• Contribute to the CaSPAR community

Overall I feel I met all my goals, albeit I am left with the impression that I could have been more proactive on collaborating with CSER people and their research.

Six months after, my main academic outputs during this period were:

• An exploration of optimal intervention timing from a decision-theoretic standpoint, plus a blogpost with a summary and discussion. This was a very ambitious project and I am glad I undertook it. But as I advanced and received very little positive feedback I resolved to publish an unpolished exploration and focus on my other projects.
• A preliminary exploration of quantum computing technologies from a philanthropic point of view. This was a blurb collecting my thoughts in the issue rather than formal academic work. Putting it out there helped me connect with a coauthor for the next paper.
• A paper forecasting developments in quantum computation I worked on with Jess Riedel. Despite many limitations we are facing, I am really proud of this work and I am hopeful we will get to publish it soon.
• A paper discussing the security consequences of quantum cryptanalysis. I am unsure of the quality of this paper, and I am currently looking to submit it somewhere where I can get more feedback.

Overall this experience has helped cement my research agenda, which is now oriented towards improving decision-oriented forecasting and understanding the impact of emergent technologies.

Beyond academic writing, I accomplished a lot in other projects:

• I wrote a blogpost about heavy tail distributions, that led me to research more in depth the two central theorems of extreme value theory.
• I peer reviewed an article for the journal Futures.
• I wrote an article with Alex Hill with a preliminary approach to understanding the historical importance of research in moral philosophy.
• I wrote another article with Alex Hill on epidemiological modelling and social distancing following the COVID19 UK outbreak, that we ended up deciding not to publish as we found a mistake, and subsequently we decided were quite confused about the topic and did not want to contribute to a wave of bad epidemiology nor waste valuable time of domain experts we could ask for feedback.
• I published a divulgation piece in Spanish on Nvision about existential risk, focusing on Nuclear Risk and Artificial Intelligence.
• I gave a talk in Spain at Sngular about existential risks.
• I published a summary and review in Spanish of Toby Ord’s new book.
• I gave three internal talks at CSER explaining my work.
• I hosted two workshops at CSER on forecasting.
• I participated as an expert on a Delphi study on forecasting led by Ross Gruetzemacher.
• I organized and hosted two remote undergraduate workshops on generalist research as a collaboration between CSER and CaSPAR. Props to Jess Whittlestone and Ashwin Acharya who were instrumental in making these happen.
• I conducted ~7 interviews for a follow up study of the impact of CaSPAR 2019.
• I received at the CSER office and mentored about ~10 Cambridge undergraduates on career decisions and research.
• I self studied a lot of statistics (especially causal inference and bootstrapping) and machine learning (specially reinforcement learning).
• I attended the International Conference on Learning Representations (ICLR) 2020.

In terms of career planning, during this period I applied to ~5 jobs and ~10 graduate studying programs. In the end, I was awarded a Marie Curie grant to study a PhD in Bayesian Reasoning at Aberdeen, which I will start in July.

Meanwhile, I have accepted a commision from the Digital Future Society to be the lead author on a report for the Spanish government on Artificial Intelligence and Climate Change, which I will be working on for the next two months.

Conclusion

My experience receiving the CLR grant, visiting CSER and living in Cambridge was very enriching, albeit challenging at times.

Even though it did not feel like that while I was doing it, now looking back on these months I realize I have not only met but surpassed my goals.

I’ve come off with a very positive view of CLR and CSER, and I would encourage people to check them out and support their work.