Research

AI-augmented research on frontier science, autonomous research systems, and the economic consequences of machine intelligence.

mutome

An autonomous research lab for frontier science and engineering.

Mutome runs many candidate research routes in parallel, then promotes only replayable evidence: protocols, code changes, proof routes, metrics, artifacts, and recorded decisions.

ProposeVerifyIterate

Research theses

01

The polymath becomes operational again

AI lowers the cost of entering new fields. It does not create lone geniuses, but it lets disciplined researchers hold many provisional maps at once and test bridges across domains.

Read the essay
02

Production becomes cheap; verification becomes central

In software, agents increasingly write and review code while humans specify, audit, and decide. Science faces the same shift: if agents can propose discoveries, the scarce layer is proof, reproducibility, attribution, and judgment.

03

Research needs new trust infrastructure

The hard question is not whether an AI can produce plausible claims. It is how we know which claims are interesting, which proofs are correct, which experiments are real, and which results deserve attention.

How we work

Explore widely

Models help map literatures, compare frameworks, draft code, and surface cross-domain analogies worth testing.

Verify narrowly

Promising claims are pushed through local tests, reproducibility checks, solver traces, formal tools, or expert review.

Keep memory

Useful failures, contradictions, negative evidence, and source trails are archived so later work compounds instead of resets.

Current areas

Autonomous research systems

Eval-driven lab loops, candidate archives, benchmark adapters, and hardening protocols for AI-assisted discovery.

Cosmic topology and CMB structure

Compact-universe models, low-ell CMB correlations, and falsifiable topology tests.

Holography and black-hole information

Island formulas, Page curves, and entanglement bounds on compact spaces.

Casimir energy and dark-energy tests

Vacuum-energy calculations, holographic dark energy, and public cosmology data.

Extremal combinatorics

Frankl-type union-closed set problems, entropy barriers, and proof reductions.

Scattering geometry and entanglement

Amplituhedron, celestial holography, and positive-geometry bridges.

Formal verification pipelines

Lean, PB certificates, solver traces, reproducibility, and expert review gates.

Public work

We use AI systems for breadth, literature review, coding, simulation, candidate generation, and critique. Claims are promoted only through evidence gates: reproducible artifacts, tests, formal checks where possible, and human review. Public work shows the direction; not every internal workflow detail is disclosed.