Project Goal

Project Goal

The goal of this project is to move understanding forward such that radical AI alignment is possible. The most ambitious and important kind of alignment is not how to match human ideals. It is to create a system that agentically maximizes for definitional good.

Therefore, we aim to develop a comprehensive philosophical framework in which questions like 'what is consciousness?', 'what does this [specified system] feel and sense?', 'what can we know a priori about what exists?', etc. can be answered.

This is a difficult goal to achieve, but it's easier when not bogged down by discourse on irrelevant problems and/or notions we reject. We take some positions (often: more like intuitions) as true or guiding principles.

We hold the following positions:

- Reality is identical with logic: "What is rational is real; and what is real is rational" - Extending that, cognition is best understood through symbolic terms. Martin-Lof Type Theory

 provides a basis that makes legibility possible, makes mental processes intuitive,
 allows us to natively speculate about importing and adjusting algorithms for any core tasks,
 makes proof driven AI alignment tractable, etc.

- Utilitarianism is the closest to a coherent ethical position.

 - Rightness and wrongness are 'objectively' true only as valence,
   which is a section of reality that states a judgment about itself.

- Time is emergent, reality is best viewed as a tenseless object. - Wolfram's Ruliad and Max Tegmark's MUH are close to the truth but require better

 justification.

- Open individualism is the closest to the truth wrt personal identity. - Anthropic reasoning and the simulation hypothesis are perfectly valid and natural.

Immutable context document supplied to every agent turn.