Deontological (rule-based) ethics are often implicitly justified by the good outcomes their rules are presumed to create. If a moral rule was known to produce the worst possible results, its proponents would likely abandon it, revealing a hidden consequentialist foundation for their beliefs.

For difficult decisions, ask the simple question: "What does right look like?" and then do that. This framework simplifies complexity. While doing the right thing can be harder or more expensive in the short term, it consistently leads to better outcomes in the long run.

Shifting from a black-and-white "right vs. wrong" mindset to a probabilistic one (e.g., "I'm 80% sure") reduces personal attachment to ideas. This makes group discussions more fluid and productive, as people become more open to considering alternative viewpoints they might otherwise dismiss.

When facing a difficult choice that creates persistent unease or uncertainty, it's often a signal that the correct path is to decline or opt out. This heuristic, borrowed from investor Naval Ravikant, helps cut through complex analysis paralysis, especially in situations with ethical ambiguity.

Aligning AI with a specific ethical framework is fraught with disagreement. A better target is "human flourishing," as there is broader consensus on its fundamental components like health, family, and education, providing a more robust and universal goal for AGI.

When faced with imperfect choices, treat the decision like a standardized test question: gather the best available information and choose the option you believe is the *most* correct, even if it's not perfect. This mindset accepts ambiguity and focuses on making the best possible choice in the moment.

For robust, high-stakes grantmaking, separate analysis into three layers. First, empirical uncertainty (what will happen?). Second, normative uncertainty (what outcomes are most valuable?). Third, meta-normative uncertainty (how should we aggregate different moral views and risk preferences?). This framework clarifies thinking.

Under the theory of emotivism, many heated moral debates are not about conflicting fundamental values but rather disagreements over facts. For instance, in a gun control debate, both sides may share the value of 'boo innocent people dying' but disagree on the factual question of which policies will best achieve that outcome.

Rather than relying on a single AI, an agentic system should use multiple, different AI models (e.g., auditor, tester, coder). By forcing these independent agents to agree, the system can catch malicious or erroneous behavior from a single misaligned model.