Gage Siebert
PhD candidate, Arizona State University · algorithmic complexity → AI safety
I'm a PhD candidate at Arizona State University, working with Sara Walker on applied algorithmic complexity. My thesis involves formalizing interpretable, substrate-specific languages of computation, building software that makes complexity-theoretic analyses practical, and using these tools to find hierarchical structure in biological systems.
After my PhD I want to work on AI safety, particularly intepretability. I think this is among the most important problems to work on right now. Much of my PhD research has addressed the question: given a system, how can we build a computational model of it that is interpretable. I've spent my PhD building formal tools to answer that for chemical and biological systems, and I'd like to bring the same approach to neural networks.
Before ASU, I studied physics and math at UW–Madison, where I worked on radio cosmology, millisecond pulsars, the optics of a NASA balloon microwave telescope, and the origin of life.
Available for AI safety roles from April 2027.
Currently: Finishing PhD and working through ARENA curriculum.
Selected papers
- Assembly Spaces and Formal Grammars: a General Correspondence and Fast Bounds on Assembly Indices In Revision
- Practical Assembly Theory, Including Protocols, and a Python Toolkit under review
- Rapid Exploration of the Assembly Chemical Space of Molecular Graphs
All publications, talks and honors →
Elsewhere
I also like making fractals, taking astrophotos and writing the occasional short story; they're collected on the miscellaneous page. A good place to start is The Oracle's Fractals, a short essay about a class of fractals that contain every computation.