> whoami
UtilityHotbar
About me
- I have been coding since ~2018. Email me here: mutatismutandisetplusultra (the usual symbol) gmail (another symbol) com. Book a meeting here: https://calendly.com/mutatismutandisetplusultra/30min (If you don't know me personally, please send an email first! I do not bite and actually respond to emails, I promise)
- I program simulations, tools, toy compilers/languages, games, and multimedia projects/ARGs
- My research interests include active inference, fundamental theories of learning, the origins of life, AI safety/interpretability (what does it mean for AI to "go well"?), the origin and organisation of human knowledge domains, and the institutional incentives for developing AI technology.
Research
- The Human Inductive Bias Project - I am currently organising and running a research project on the intersection between active inference and machine learning, covering learning dynamics, fundamental research, and possible future architectures. The motivation is to apply these insights to AI safety, in particular looking at how and why capabilities are acquired and how generalisation occurs.
- Harms from Increasingly Agentic Systems - A paper I was a co-author on, talking about the risks from AI systems we are yielding autonomy to.
- AI Generated Multimodal Disinformation Networks - Conference presentation demonstrating some of the capabilities for AI to be used for complex disinformation narratives.
Writing
- Promathematheon - An attempt to derive a coherent philosophy around the origin of life, intelligent behaviour, high level organisation, human intelligence, artificial intelligence, and what all of this means for our present condition.
- A Clever Name Goes Here - A small substack for occasional pieces related to AI safety, philosophy etc. Selected post: A Letter to His Highness Louis XV, the King of France.
- Turning 22 in the Pre-Apocalypse - A collection of essays as I face the end of my 21st year on this planet. Covers and questions rationality, AI development, the Humanities, and my own motivations for doing this.
- The Savage Computers - A modern web novel about AI, climate change, and the present state of the world. Includes interactive and non-linear elements. Shortlisted for the UK's New Media Writing Prize Student Prize.
- Automatic Memory - Issue N - A zine made by me (graphic design + words).
Programming Projects and Sites
- Socratides - A demo for a narrow AI system that generates human understandable and editable knowledge graphs out of texts that can be pruned and refined to enhance textual analysis. Source code here.
- Zhengzi Archive - Online facing part of an ARG I ran.
- Civsim - Automated generation of civilisations and simulation of their growth and decay.
- Journey - Automated D&D questing.
- Shellsim - A self-contained simulation of a bash like shell: piping, streaming, internal filesystem
Games
Minor Projects
- Memex Note - An experimental text editor based on cells in Jupyter notebooks and other funky concepts.
- Showmd - Displays markdown files in a semi-pretty way in the shell.
- ARG Tools - Tools for running and making ARGs.
Old Stuff
- Progress Game - Short incremental game with paper.css.
- Bytedelver - Unfinished roguelike using rot.js and NES.css.
- Mass Defect - Unfinished alchemy-themed incremental game.
- Yet Another Web Toolkit - A web playground for developing games, applets, slides, or small fixed-viewport websites. Dangerously unfinished.
- oldbook.css - A CSS theme that draws from old books. For some reason, my most popular project by far. No JS needed. I'm working on a new version of this.
- neoweb.css - Another CSS theme, with a demo page. No JS needed.