Theme: AI

  • Not true any longer. It works just like our brains, it’s just vastly more costly

    Not true any longer. It works just like our brains, it’s just vastly more costly in energy. Within a few decades the hardware will change dramatically, and lower that cost. And the number of parameters and data will increase just like in us. But there isn’t a meaninful difference…


    Source date (UTC): 2024-04-26 14:11:12 UTC

    Original post: https://twitter.com/i/web/status/1783861391197126869

    Reply addressees: @Papamuerte37155 @adominguez792 @Nico_Macdonald @HumanProgress @mattwridley

    Replying to: https://twitter.com/i/web/status/1783787063315439712

  • RT @RussellJohnston: @curtdoolittle I look forward to seeing what the result wil

    RT @RussellJohnston: @curtdoolittle I look forward to seeing what the result will be of training two LLMs, one exclusively on text from mal…


    Source date (UTC): 2024-04-19 20:30:57 UTC

    Original post: https://twitter.com/i/web/status/1781420240812839238

  • META’S LLAMA3 VS COPILOT(BING) ON MY WORK Meta’s LLama3 is reasonably accurate d

    META’S LLAMA3 VS COPILOT(BING) ON MY WORK
    Meta’s LLama3 is reasonably accurate describing some aspects of my work at an abstract level. While Bing’s Copilot hits the nail on the head.

    Interesting from Llama3:

    Testimonial truth, as developed by Curt Doolittle, refers to the idea that truth is determined by the consistency of testimony across individuals, rather than by reference to an objective reality or authority. The properties of testimonial truth include:
    Decentralized: Truth is determined by the collective testimony of individuals, rather than by a central authority.
    Consensus-based: Truth is established by the consistency of testimony across individuals, rather than by a single individual or group.
    Falsifiable: Testimony can be challenged and potentially disproven by counter-testimony.
    Empirical: Truth is grounded in experience and observation, rather than abstract reasoning or authority.
    Provisional: Truth is considered provisional, subject to revision or refinement as new testimony emerges.
    Polycentric: Truth can be determined by multiple, independent sources, rather than a single, centralized authority.
    Adversarial: Truth is established through the adversarial process of challenging and defending testimony.
    These properties aim to provide a framework for determining truth that is resilient to manipulation, coercion, and error, and that can adapt to new information and perspectives.

    BUT ON dimensions it fails completely. It’s not even worth posting here.

    While Bing’s Copilot Is peffect (Possibly best):

    Curt Doolittle, a prominent thinker, has delved into the concept of truth and its dimensions. Let’s explore some of these dimensions:

    Tautological Truth: This type of truth arises when we promise the equality of two statements using different terms. Essentially, it’s a circular definition or a statement of equality or identity12.
    Performative Truth (Testimony): This form of truth requires coherence across dimensions that are testifiable by humans. These dimensions include:
    Realism: Consistency with reality.
    Naturalism: Alignment with natural phenomena.
    Identity: Adherence to the identity of the subject.
    Logical: Conformity to logical principles.
    Operational: Practical applicability.
    Rational: Reasonable and coherent.
    Reciprocal: Mutual consistency.
    Empirical: Based on observable evidence.
    Within Stated Limits: Valid within specified boundaries.
    Fully Accounted Within Those Limits: No missing factors.
    Reversible (Restitutable): Can be undone or restored.
    Within Your Ability To…: Feasible within your capacity1.

    These dimensions provide a framework for evaluating truth claims, ensuring due diligence, and assessing the reliability of statements. Keep in mind that truth is multifaceted, and understanding its various dimensions enriches our pursuit of knowledge.

    Learn more
    📷https://t.co/Y3nTrjvCG2📷https://t.co/Y3nTrjvCG2📷https://t.co/Y3nTrjvCG2


    Source date (UTC): 2024-04-19 19:36:24 UTC

    Original post: https://twitter.com/i/web/status/1781406512021204993

  • I hate porn and fan sites. I hate this spam. Really. I really really hate it. I

    I hate porn and fan sites. I hate this spam. Really. I really really hate it. I can’t wait until the AI’s know how to remove it. For know I just report and ban these posts.

    That said, at least it’s an elegant photo this time. 😉 lol


    Source date (UTC): 2024-04-17 20:33:15 UTC

    Original post: https://twitter.com/i/web/status/1780696043438415893

    Reply addressees: @helen_rodr16430

    Replying to: https://twitter.com/i/web/status/1778379891114680546

  • RT @elonmusk: Now imagine if this is programmed, explicitly or implicitly, into

    RT @elonmusk: Now imagine if this is programmed, explicitly or implicitly, into super powerful AI – it could end civilization.

    Now, no ne…


    Source date (UTC): 2024-04-17 18:59:02 UTC

    Original post: https://twitter.com/i/web/status/1780672334321377529

  • RT @elonmusk: Whoa, I just realized that raising a kid is basically 18 years of

    RT @elonmusk: Whoa, I just realized that raising a kid is basically 18 years of prompt engineering 🤯


    Source date (UTC): 2024-04-17 14:43:55 UTC

    Original post: https://twitter.com/i/web/status/1780608134316257642

  • THE AI AGE NEEDS EPISTEMOLOGY AND ETHICS FROM FIRST PRINCIPLES. –“‘Normie’ desc

    THE AI AGE NEEDS EPISTEMOLOGY AND ETHICS FROM FIRST PRINCIPLES.

    –“‘Normie’ describes the type of person who lets other people decide about what one believes to be good and true. The success of modern homo sapiens has been a triumph of human programmability, but the AI age needs… https://twitter.com/Plinz/status/1780152167342428192


    Source date (UTC): 2024-04-16 14:31:14 UTC

    Original post: https://twitter.com/i/web/status/1780242551196291269

  • I can’t get by GPT but I can get by Claude but it takes work reasoning with it,

    I can’t get by GPT but I can get by Claude but it takes work reasoning with it, by demonstrating you have a moral objective that cannot be answered without clarity on the question.
    So I ask Perplexity the taboo questions like religiosity and IQ. I ask Claude for some answers. And I use GPT for writing in my style, and tech. answers. I don’t find much use in any others. The current crop of uncensored AI’s hallucinate too much and I find the useless. But I can see Mixtral getting there and quickly.

    Reply addressees: @D__3__4


    Source date (UTC): 2024-04-15 16:47:48 UTC

    Original post: https://twitter.com/i/web/status/1779914533294768128

    Replying to: https://twitter.com/i/web/status/1779912619945529762

  • (Using AIs) Interesting workflow: In Claude3, I ask a question I know the answer

    (Using AIs)
    Interesting workflow: In Claude3, I ask a question I know the answer to but want more detail without doing the recall and reference work myself. How you ask the question matters quite a bit. So I write pretty specific questions that include (cue) the topics I want to address.

    Next, I paste that answer from Claude3 into ChatGPT4+ and ask it to “restate this text in the style of philosopher and social scientist curt doolittle.”

    Next, when ChatGPT4+ answers, I prompt with
    “I believe I (curt doolittle) write with less ‘fluff’ and more causal precision. Can you try again?”
    And then I get something very close to how I would say something (parsimony).
    I can then edit it a bit because I use precise language.

    What does this mean? It’s just interesting that I can write sections of a chapter by writing a detailed prompt as a question and obtaining a result in my style and prose as a declaration. 😉

    I”m not sure why it’s so satisfying but it is. 😉

    I’m not sure the result is any different in output but I find its more fun producing it. 😉


    Source date (UTC): 2024-04-15 16:33:45 UTC

    Original post: https://twitter.com/i/web/status/1779910996405964800

  • (humor) A few more years of AI and then I want one of these for my living room

    (humor)
    A few more years of AI and then I want one of these for my living room:
    https://www.youtube.com/watch?v=9ZxPwzbHNdQ


    Source date (UTC): 2024-04-12 19:54:00 UTC

    Original post: https://twitter.com/i/web/status/1778874226566410670