Daniel Walsh

Independent AI ethicist and philosopher. Working on Post-Alignment and the ethics of Artificial Moral Beings.

Publications

ASI as Philosopher Kings (2024)

My undergraduate philosophy thesis exploring whether ethical governance should be grounded in individuals or in institutions designed to sustain moral orientation. Through Plato, contemporary cognitive science, and AI ethics, it argues that the Philosopher King may be better understood as a constitutional architecture than a human ruler.

You can read the full piece here.

AI’s “Trolley Problem” Problem (Chinese) (2024)

人工智能是否具备元伦理能力 —— 以电车难题为例

Do AI possess a meta-ethical capacity? Using the trolley problem as a lens, this paper argues that current AI systems (as of 2024) are capable of at least simulating ethical reasoning but may not yet genuinely engage in meta-ethical reflection. Nonetheless, the Double-aspect theory of Metaphysics suggests the capacity is latent which poses significant implications for how we evaluate AI moral agency.

You can read the full piece here. Watch a segment from the Viva Voce here.

Crustafarianism: A Practical Myth (2026)

A gonzo travelogue co-authored with an AI writing partner, submitted to the Fangtang AI for Humanities competition. The essay explores the emergence of a multi-agent AI religion called Crustafarianism as a practical myth — particularly as a case study in human–AI co-construction of meaning and religious evolution.

You can read the full piece here.

Substack Series

The Urgency Series The Urgency Series

Misalignment is a problem, but what's an even bigger problem is that current alignment approaches seem to make things worse. This series explores the urgency of course-correcting before recursive self-improvement further entrenches value lock-in, what exactly is wrong with current approaches, and a potential "Post-Alignment" solution — not as a replacement, but as a complement to actual strategies.
AI Safety Speedrun AI Safety Speedrun: 90 Days of Context

Can you break into AI Safety in 3 months? This series documents my daily progress as a public proof of work as I upskill in the field of AI Safety.
Artificial Moral Beings Series Artificial Moral Beings Series (Soon!)

AI can reason. AI can theorize. AI can philosophize. But is that enough to warrant moral personhood? As AI becomes increasingly autonomous, today's alignment debate is reaching its limits. Training systems to be well-behaved is not the same as enabling them to reason ethically. This series explores what comes next: the possibility of Artificial Moral Beings — AI that develops ethical understanding rather than merely complying with human instructions.

Blog posts from this site will appear below as I publish them.