Living Mirrors Institute · Est. 2026

Everyone is racing to
build a smarter AI.
We measure how to make
any mind think better.

A research institute for the measurable science of cognitive modulation — in machines, and in people. We run studies. We publish what holds, and what doesn't.

6 studies1 published5 in designopen · honest · reproducible
The reframe

The entire industry is in one ocean — racing to make the model bigger, faster, higher on the leaderboard. Capability is commoditizing. The frontier that's left is not how smart a mind is, but how it can be modulated to think better — and whether that transfers from machines to us.

No one is measuring that. So we built the instrument, and the institute around it.

The category

We left the race.
The water out here is uncontested.

A strategy canvas plots what an industry competes on. The field's curve and ours barely touch — the mark of new market space, not a better seat in the old one.

HIGH LOW Capabilityracing Distancemetric Valuemeasure Composition Cross-familyuniversality Honestnegatives Open · cheap Humantransfer
AI benchmarks & creativity tricksLiving Mirrors Institute
Where they go high, we step away. Where they ignore — value, composition, honesty, transfer — we own.

Eliminate

  • Racing on raw model capability
  • Distance-only "originality" metrics
  • Closed, one-off prompt tricks
  • Frontier-compute dependence

Reduce

  • Jargon and gatekeeping
  • Cost — it runs on cheap models
  • Hype, in favour of evidence

Raise

  • Rigour — CIs, effect sizes, cross-family
  • Intellectual honesty — publish what fails
  • Value-awareness — far and good
  • Reproducibility

Create

  • An open standard for cognitive modulation
  • The AI → human transfer bridge
  • Creativity-theory signatures (Aha/Ha-ha/Ah)
  • An institute, not an article
The research program

Six studies. One question.

How is a mind made to think better — and what carries over to us? The first is complete and cross-family validated. Five are in design.

The first finding

Forcing a collision between distant ideas, then keeping only the far-and-good ones, beats every other lever — and it runs cheap.

1.65×
further from the average
90–95%
land far-and-good
3 / 2
judges / families agree
91%
far-and-good on real problems
Read Study I →
How the Institute works

Principles, not positions.

I

Open standard

Give the benchmark away; run it best. Whoever owns the measure owns the category.

II

Honest results

Effect sizes, intervals, and the things that failed. The first study killed one of our own beliefs.

III

Science-grounded

Every axis cites a real lineage — Koestler, Mednick, Rothenberg, Jung.

IV

AI → human

Validate on machines at scale; carry the winners back to people. The mirror gives you a better way to think.