Hello. I’m James, a mathematics student at the College of William & Mary.
I’m interested in AI safety research, particularly technical governance and AI control. I also want to learn more about the large scale social and economic consequences of building human-level artificla intelligence, in order to help the world prepare.
I most recently worked as a Troy A. Cullen Emerging Technology Fellow at the NukeLab, investigating the potential near-term effects of AI on nuclear proliferation and command.
I previously worked at the World Bank and Global Environment Facility studying the socioeconomic impacts of their environmental protection projects.
AI progress over the last 3 years has been incredibly swift. In 2022 OpenAI released ChatGPT, the world’s first consumer product demonstrating a command of natural language comparable (but still inferior to) humans.
Previous models like GPT-2 had shown the ability to string together words into complete and somewhat coherent sentences, but nothing like an AI-powered chatbot that could have a somewhat intelligent conversation about nearly anything.
I’ve made some updates to my website and want to catalog them here. Previously I had a landing page built with Next.js linking out to my GitHub profile and social profiles. This was fine but using Next.js and Typescript for my needs felt like overkill, and I wanted to be able to extend the site more easily. So this new version is built with the static site generator Hugo and uses bare HTML/CSS. I managed to make the initial version without any Javascript, although that may change in the future.