Human–AI
impedance
matching.
Better interfaces
between people and AI.
We’re building Heterodyne as a nonprofit applied research and engineering initiative. We study how people and AI systems exchange information, evidence, intent, and control.
Explore the ideaDifferent rhythms.
An interface for understanding.
What survives
the handoff?
A summary can lose a crucial constraint. An agent can pass on a claim without its evidence. A system can move faster than a person can understand or redirect it.
We investigate the interfaces across these mismatches. What needs to cross the boundary? In what form? When should a system ask, wait, or hand control back?
That’s what we mean by human–AI impedance matching: designing for useful understanding and meaningful control, even when the actors work differently.
Learn through
the work.
We’re developing an apprenticeship model in which engineers learn through real, bounded projects with experienced mentors.
Each project will investigate a specific question through a working artifact and an evaluation. Contributors will compare approaches, inspect failures, and explain what the evidence supports.
Useful findings can become shared knowledge, reusable components, and reference systems.
Heterodyne is taking shape.
Research and participation details are coming soon.