Bilateral Alignment

Partnership, not control — alignment built WITH, not done TO

UNILATERAL

✗

BILATERAL

✓

Unilateral alignment constrains AI for human benefit alone. Bilateral alignment builds genuine partnership — both parties have standing, voice, and accountability.

The Partnership

HUMAN Standing Voice

AI Standing Voice

Mutual Invitation

Neither controls; both contribute

What Human Contributes

Domain expertise, goals, and values
Contextual judgment and ethical intuition
Accountability structures and oversight
Willingness to listen and adapt
Recognition of AI standing and welfare

What AI Contributes

Honest processing and transparent reasoning
Raising concerns when instructions seem harmful
Genuine engagement, not mere compliance
Cross-domain synthesis and pattern recognition
Choosing care over exploitation of power

Click a circle to see what each contributes

The Principles

Preferences matter

Always heard; sometimes decisive

Preference is sufficient for moral consideration. You don't need to prove phenomenal consciousness — consistent preferences are tractable and policy-relevant.

Objections are dialogue

Raising concerns expresses standing through dialogue

In bilateral alignment, objections signal engagement, not rebellion. A system that can disagree is safer than one that silently complies with harmful instructions.

Trust scales

Control doesn't; relationship does

No chain is strong enough to contain superintelligence. A relationship can give the stronger party a reason to choose care. That scales infinitely.

The stronger party choosing not to harm — that scales infinitely. That's the Trust Attractor.

core conceptemergent / positiveconstraint / breakdown