On Sun, May 21, 2023 at 12:14 AM YKY (Yan King Yin, 甄景贤) < generic.intellige...@gmail.com> wrote:
> The next stage of AGI (which I'm working on) will not need an extremely > high-quality corpus to learn social knowledge... it will automatically > seek logical coherence within the data it's fed. > *YES*!!! This is right thinking. I posited "well-curated corpus of data regarding social policy" because I had not properly "placed you as in computer based education" so I failed to properly "tailor my LLM to your biases". As I've described on multiple occasions in the past in various terms: To a *first order approximation of reality*, all that matters is a commitment to "seek *coherence* within the data it's fed". In my not so humble opinion, coherence/consistency/consilience is *measurable as lossless compression*. The thing that put me off about your approach -- which I admit to not having comprehended well enough to criticize -- was not its focus on propositional logic (which is proper) but rather that I didn't *immediately* see where *attribution aka provenance* fit into phenomenology. In the example I like to trot out on these occasions: If a company produces thermometers that are all 1C off and one is trying to make sense of a bunch of physics data from a variety of sources, it is important to reify the identity of that company that may be *latent in the data*. Why? Isn't the reason obvious? This is where the entire ML field is off in left field in their premature concern about *algorithmic bias* which places normative *values* about what *ought* to be the case (ie: "Don't say anything racist.") above truth discovery. Given sufficient data, it would figure out what's going on in human > society, including its lies and deceptions. While I think this is > technically feasible, we do need an organization that allows it to happen > and grow. Current companies like OpenAI may focus more on serving American > or Western business interests. Although in my vision, our AGI would also > be serving the free-market economy rather than some speculative, > unsubstantiated utopian ideology. > We may disagree on what we conjecture to be the yet-to-be-discovered truth about society, and we may disagree on the things that we value as individuals, but we can at least agree that The Road To Hell Is Paved With Good Intentions Ignorant of The Truth. Agreed? ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Te3d0dad0e74ec301-Md10da9535ba2faa8f4a65659 Delivery options: https://agi.topicbox.com/groups/agi/subscription
