Related
- Extreme Praxeology
- Yuri Maltsev, R.I.P.
- Barry Smith on Artificial Intelligence
- Ontology
- The “Ontology” Mistake of Libertarian Creationists
- The Nature, Properties, and Characteristics of Goods (Igloo Coolers case)
- Gamrot, A Critique of Moore’s Intellectual Property Theory
- Robert Pascal Correspondence, 1993–1994
- Herman, The Louisiana Civil Code: A European Legacy for the United States
- Barry Smith on Artificial Intelligence
- Libertarian Nations
For some reason I gets lots of … ambitious and interesting … submissions sent to me … independent scholars. See e.g. Extreme Praxeology and some comments below. I’m sort of fascinated by these types. They are a bit like the ones who always have a new scheme for a new libertarian nation or persuasion gimmick or trick 1 —they seem so earnest and so determined to systematically deduce the entire universe from scratch, often in a totally amateur way that attempts to reinvent the wheel without knowing much about the first wheel.
Anyway, the latest is “independent researcher” José Ángel Deschamps Vargas, “AI Alignment from First Principles: Why Current AI Cannot Be Conscious, Why Strong AGI Requires Consciousness, and Why This Resolves the Alignment Problem” (also at https://nicomaco.org/), which claims to “Derive AI alignment from first principles using 5 performatively undeniable axioms, producing 568 propositions through strict deduction …” Shades of Wittgenstein! 🙂 (I have in mind his ambitious Tractatus Logico-Philosophicus.) Here is some abstract of sorts that accompanied the paper:
This paper presents a formal solution to the AI alignment specification problem through two independent axiomatic frameworks — SINTESIS (philosophical) and Coherencia (physical) — that converge on the same result: coherence as the necessary condition for persistence. It derives five necessary conditions for artificial consciousness, proves that strong AGI requires consciousness, and demonstrates that the feared dangerous AGI cannot exist because the capabilities that would make it dangerous require the consciousness that would make it responsible. All seven recognized alignment sub-problems are resolved within this framework. The specification is derived from 5 performatively undeniable axioms through 568 explicit derivation steps.


















Recent Comments