Applying to Oxford DPhil, deploying INFLOW-AI, and writing abstracts for the Think7.

Goals for the Year

As my work on the INFLOW project finally wraps up, I’ve finally had time to think about what exactly I want to achieve this year. Most saliently, I want to put my ideas into action and actively found a research group to focus on addressing algorithmic governance challenges faced by public sector organisations (e.g. implementing automations, machine learning forecasts, etc.). I think last year I did a fantastic job of achieving this on my own, but with all these projects under my belt (and finally some spare time), it’s finally time to scale up. I’ve written a list of what I need to make this achievable (in no specific order): (1) acquire an office for project work, (2) acquire a co-founder, (3) acquire a board of directors, (4) start a formal process to acquire student volunteers. Fortunately I have no need to acquire projects, as I already have a backlog of requests I need to attend to… (with more likely incoming).

I also want to hopefully become more punctual about writing these updates! (I may or may not be writing this January post in May of 2025…)

Think7 Abstracts

International governance may be significantly more idealistic than effective, but if there is any chance that brining international attention to critical issues actually contributes to change, I want to be involved. With this mindset, I submitted three (which is a lot) abstracts to the Think7 for the Canadian G7 Summit this year. I managed to scrape together an entire cadre of expert co-authors from around the world in a very short period of time to make the tight deadlines with three goals: (1) to reduce market concentration in the foundation model market, (2) to argue for governments to make policy more effective by adopting Rules as Code, and (3) to reduce beef consumption by removing national subsidies on for beef farming (this is the longest shot of the three, as it is the least aligned with current Summit priorities, despite annual lip service to environmental causes).

The goal is to increase international cooperation to prevent monopolies from dominating as foundation models become increasingly economically important.

Writing Rules as Code is a highly cost effective way to increase public access to complicated legislation (e.g. social benefits) and simulate policies before they are implemented.

Arguably the most important abstract, which I submitted to both the T7 and T20.

Are international forums like the G7 actually effective in shifting the direction of national policy? Time will tell for these three proposals.

Other Work

I fully deployed the INFLOW-AI model (available on GitHub here). This code will soon be run through JASMIN, meaning that the predictions will be live and accessible to humanitarian agencies actively using the model to make funding and deployment decisions!

This month I was also selected to be a “Young AI Leader” with the ITU’s AI for Good program. Essentially they want me to recruit young people in Oxford interested in contributing to AI for Good, but the benefits to doing so in terms of actual AI for Good are currently unclear to me. I think they maybe just want access to a large network of people .

Oh, and I spontaneously decided to apply for the Oxford DPhil in Statistics. Updates to follow.

We Did a Cruise

We did a cruise, courtesy of Alec’s parents. Cruising would certainly not be my first choice for vacationing given the environmental and local economic impacts of cruises, but it’s really difficult to look a gift cruise in the mouse (not to mention this would be my first time meeting Alec’s parents)

It seems every European city with any amount of tourists has an Iris Store. This is a weird shop that sells you a picture of your own Iris, blown up to wall painting size. They are clearly targeted towards tourists, but why would anyone think that bringing home a picture of your own eye (which is with you always) is a good memory of travel? To further the mystery, these shops are almost always empty with 2-4 bored employees sitting inside staring blankly at the hordes of uninterested shoppers passing them by. What is the business model? How is this profitable? It’s certainly and intriguing mystery.

If you’re travelling around Europe, it’s a fun game to try and spot the Iris Store in the cities you travel to.

Key to our journeys was visiting as many art museums as possible. We were particularly excited to see Gentileschi’s Judith Slaying Holoferenes at the Uffizi, and made a bit of a game out of Judith-spotting at the other galleries we visited (I still think Gentileschi’s is the best).

A solid 7/10 Judith.

A rather plump and docile-loking Judith.

A very bloodless and emotionally blank Judith, just after the beheading with the gore hidden out of frame.

Caravaggio’s take on the subject, clearly a major inspiration for Gentileschi.

Here’s our girl! Super dynamic and gory with parabolic blood arcs, a relatively novel technique for the time.

We found this little store in Marseille.

Apparently Pisa-leaning on things that are not the Tower of Pisa really confuses locals.

Travelling to French coastal cities in January is a bit strange, as everything is closed and there are no people.

I found the Sagrada Familia to be more interesting on the inside than the outside (which is usually all that people think of).

Barcelona’s excellent urban planning and beautiful superillas can be credited to 19th century urban planner Ildefons Cerdà.

Us.

Dream bigger.

Other Life

Someone stole my bike wheel while we were away…

Home-made Ukrainian breakfast with Oxford friends.

Celebrating Lunar New Year at Mansfield College with forks and knives.

Complaining about Ea-Nasir’s copper shipment with a friend visiting from Toronto.

Built some Lego flowers for our room – a Christmas gift from Alec’s mom.

Finally got my annual hair cut.

As a note, I’m a bit worried that these updates are disproportionately focussing on non-work things that I am doing, but those seem to be the most photogenic…