danbri: Assume most people saw this, but just in case. Nicely written paper too.
Idmclean: an argument for the adoption or exploration of paraconsistent systems of reasoning in formal methods. Final draft intended for publishing on Radical Computing and Arxiv.
Idmclean: Looking for peers to review, comment, and critique.
Idmclean: Open Science Notebook, full content. Licensed under Creative Commons Share-Alike 3.0 unported license.
DanC_: not very persuasive, to me. Why bother with symbolic methods for stuff like "people lie" when statistical methods do so well?
DanC_: I tend to agree with Chaitin that the main result of logic is computation, not modeling how the human brain works
Idmclean: Regarding Chaitin and modeling a mind whether human or otherwise: "Intelligence is nature’s AI. Thus, if AI were impossible, I would be impossible, too. Since I am possible and [Intelligence] is possible, so is AI, period. The artificial/natural distinction is an arbitrary one."
Idmclean: Contemporary statistical methods are a consequence of consistent logical modeling; they are fundamentally unable to satisfy the Liar's Paradox. No consistent deductive or inductive statistical method can decide the Liar's Paradox or a Liar's Sentence.
DanC_: my point is not that AI is impossible; it's just that symbolic approaches seem clearly inferior to statistical approaches for mushy stuff like human intelligence.
Idmclean: It seems that the paper should include a section about how fuzzy, multivalue, and statistical methods relate to paraconsistent systems. A paraconsistent system as described by Zizzi is a generalization of a fuzzy system and statistical in a similar sense as a quantum mechanical system.
Idmclean: Looking for peers to review, comment, and critique.
Idmclean: Open Science Notebook, full content. Licensed under Creative Commons Share-Alike 3.0 unported license.
DanC_: not very persuasive, to me. Why bother with symbolic methods for stuff like "people lie" when statistical methods do so well?
DanC_: I tend to agree with Chaitin that the main result of logic is computation, not modeling how the human brain works
Idmclean: Regarding Chaitin and modeling a mind whether human or otherwise: "Intelligence is nature’s AI. Thus, if AI were impossible, I would be impossible, too. Since I am possible and [Intelligence] is possible, so is AI, period. The artificial/natural distinction is an arbitrary one."
Idmclean: Contemporary statistical methods are a consequence of consistent logical modeling; they are fundamentally unable to satisfy the Liar's Paradox. No consistent deductive or inductive statistical method can decide the Liar's Paradox or a Liar's Sentence.
DanC_: my point is not that AI is impossible; it's just that symbolic approaches seem clearly inferior to statistical approaches for mushy stuff like human intelligence.
Idmclean: It seems that the paper should include a section about how fuzzy, multivalue, and statistical methods relate to paraconsistent systems. A paraconsistent system as described by Zizzi is a generalization of a fuzzy system and statistical in a similar sense as a quantum mechanical system.