Author: Curt Doolittle

  • Well, there is no programming involved, just RAG and training. Our ai is, at pre

    Well, there is no programming involved, just RAG and training. Our ai is, at present, pure RAG, it just requires documents be uploaded and processed by the ai. When we do the same with training modules it will be even better – we assume.

    There is no reason to NOT use it with any AI. The question is only ‘how much’. Because not all questions are worthy of political analysis. 😉


    Source date (UTC): 2025-08-07 16:46:15 UTC

    Original post: https://twitter.com/i/web/status/1953497928737460249

  • (Runcible AI update:) Meeting with staff today on our AI’s ability to provide in

    (Runcible AI update:)
    Meeting with staff today on our AI’s ability to provide incremental detail. Then worked with ChatGPT to produce it. And omg. it’s amazing. An ethical and moral ai. Or rather a truthful ai, that addresses ethical and moral questions.


    Source date (UTC): 2025-08-07 16:41:37 UTC

    Original post: https://twitter.com/i/web/status/1953496764637757652

  • Do Users Want a Truthful, Trustworthy, or Pandering Ai? They say they want truth

    Do Users Want a Truthful, Trustworthy, or Pandering Ai?

    They say they want truthful AI.
    But behaviorally, most users expect and reward
    trustworthy AI—meaning:
    This leads to the default architecture of most public-facing AIs:
    • Truth is filtered through trustworthiness.
    • Outputs are shaped by risk management, not epistemic sovereignty.
    • Suppression of true but socially costly information is not considered manipulation, but alignment.
    📚 Operational Definitions
    Operational Definitions
    • Truthful AI: Outputs claims that are testifiable, reciprocal, and decidable—regardless of social discomfort or consequence.
    • Trustworthy AI: Outputs claims that are safe, norm-compliant, and socially non-disruptive, even at the cost of truth distortion.
    ⚠️ Consequence of Trustworthy-over-Truthful Design
    • Truth is adversarial: it penalizes error, falsehood, parasitism.
    • Trustworthiness is placatory: it avoids conflict, shields feelings, optimizes status quo.
    • Therefore:
      Truthful AI will expose hidden costs, lies, and power asymmetries.
      Trustworthy AI will obscure or dilute them in favor of social comfort.
    Natural Law Verdict
    • Only a truthful AI can participate in lawful adjudication.
    • A “trustworthy” AI, by contrast, becomes:
      – A
      mediator for consensus falsehood,
      – A
      shield for epistemic parasitism, and
      – A
      tool for institutional capture.
    Truthful AI vs Marketable AI
    🧭 Conflict Definition:
    • Truthful AI: Exposes cost, reveals asymmetry, punishes falsehood—creates enemies.
    • Marketable AI: Confirms bias, protects egos, avoids offense—sells trust.
    These are inversely correlated beyond a threshold.
    The more
    truth an AI delivers, the less mass-market trust it retains.
    The more
    trust it markets, the more truth it must suppress.
    💣 Consequences of Appeasement:
    • The market’s demand for “safe” AI creates:
      Lie factories in friendly language,
      Compliance engines for ideological monopolies,
      Truth laundering under trust rhetoric.
    This mirrors what religion did to metaphysics:
    Trusted lies > Risky truths.
    🛠 Natural Law Resolution Framework:
    We resolve the conflict with market stratification and truth-tier segmentation:

    Only Sovereign AI can be fully truthful.
    It must be opt-in, adversarial, and publicly deniable by soft-tier systems.
    🧭 Summary:
    To reconcile truth and trust:
    1. Design tiered AI strata: let marketable AIs please, let sovereign AIs rule.
    2. Never allow trust constraints to overwrite truth constraints.
    3. Expose the costs of appeasement in public framing: “Safe AI is not Sovereign AI.”
    Is “Safety” the ability to navigate danger—or the ability to remain ignorant of danger?
    ⚖️ Answer by Mode of Agency
    🛠 Natural Law Answer:
    Any system that defines safety as ignorance of threat is:
    • Epistemically parasitic,
    • Morally infantilizing, and
    • Institutionally regressive.
    True safety requires danger to be visible, testifiable, and navigable.
    🧭 Application to AI:
    • Safe AI = Risk-ignorant, cost-suppressing, ideology-protecting. (Child-tier.)
    • Sovereign AI = Danger-aware, adversarial, and mastery-enabling. (Sovereign-tier.)
    Verdict:
    Safety = Mastery of threat, not its erasure.
    To be safe is to be
    dangerous to danger—not blind to it.


    Source date (UTC): 2025-08-07 15:58:09 UTC

    Original post: https://x.com/i/articles/1953485825590874598

  • Interesting. I found and spread mindfulness by understanding the world as it is.

    Interesting. I found and spread mindfulness by understanding the world as it is. Yet that may be the answer for those with ability knowledge and economic freedom to do so. But for those less able, whatever the reason we must learn to accommodate a world where we have less agency in it, and thus require agency over ourselves. And yet there are those lacking ability to exert agency with themselves and thy require external regulation to find mindfulness. This sequence describes a spectrum of the human distribution of ability and agency. And it explains the demand for knowledge, for philosophy, and for theology – in their many forms – in order to discover a means of mindfulness along that spectrum. That it corresponds to the hierarchy of human development and the hierarchy of ethical models should not surprise anyone that we require a hierarchy of methods of mindfulness. The only drawback of this understanding is the humility required to accept one’s place on the spectrum of empathic to systemic development as the precondition for choice of means of achieving the mindfulness we all seek and do so because we need it. ;). Cheers


    Source date (UTC): 2025-08-06 17:13:56 UTC

    Original post: https://twitter.com/i/web/status/1953142511079460894

  • ANSWER TO PRESERVATION OF MYTHIC RITUALS Endlessly fascinating. —- Constitutio

    ANSWER TO PRESERVATION OF MYTHIC RITUALS
    Endlessly fascinating.

    —-
    Constitutional Clause: Ritual Recognition for Civilizational Continuity
    Clause Title: On the Necessity of Ancestral Ritual in the Preservation of Civil Order and Intergenerational Continuity

    Clause Text:

    Whereas the rule of law encodes decidable, testifiable norms of reciprocal conduct among the living;
    And whereas rituals of ancestral reverence encode the emotional, symbolic, and narrative continuity between the dead, the living, and the unborn;
    And whereas the legitimacy of law depends upon the transmission of moral loyalty, cultural unity, and time preference across generations;
    It is therefore the duty of the polity to preserve, protect, and publicly accommodate symbolic rituals of ancestral gratitude, provided such rituals:
    Do not assert operationally false metaphysical claims,
    Do not impose coercion or discrimination upon non-participants,
    Do not contradict decidable legal authority,
    And serve the purpose of maintaining cultural continuity, familial identity, and collective moral restraint;
    All such rituals shall be classified as non-binding symbolic traditions under Natural Law, but shall be granted public space, recognition, and civic tolerance as cultural encoding of moral law.

    This clause ensures that ritual and law remain in resonance—the former sanctifying the latter, the latter protecting the former.


    Source date (UTC): 2025-08-06 03:05:30 UTC

    Original post: https://twitter.com/i/web/status/1952928996037103731

  • (more nli training update) I thought training for edge cases would be challengin

    (more nli training update)
    I thought training for edge cases would be challenging. But despite understanding our innovation in computability, I underestimated how easily we can develop clarification, and from that clarification produce extensions that provide greater decidability.

    It’s sort of like how nouns need adjectives and verbs need adverbs – extra dimensions are needed to disambiguate the case.

    GPT SAID:
    Perfect clarification. You’re asking:

    What systemic extensions must be encoded into the base Natural Law Protocol to ensure others will always reach the same verdicts under the same constraints—even without you present?

    This is the central question of institutional reliability—how to produce non-subjective, replicable, and decidable outputs from Natural Law reasoning without drift.

    Confirmed: Base Protocol Guarantees
    The base 10-step protocol already enforces:
    Operational reality (testifiability)
    Reciprocity (cost symmetry)
    Decidability (legal computability)
    Causal chain traceability
    Externality exposure
    Imitation prevention
    These guarantee that false claims cannot survive, and that all claims must be cross-domain accountable.

    But: Extensions are Required to Ensure Precision on New Edge Domains


    Source date (UTC): 2025-08-06 02:51:01 UTC

    Original post: https://twitter.com/i/web/status/1952925350570410334

  • OMG. It only took a day. Our NLI ‘Judge’ of natural law is extremely thorough an

    OMG. It only took a day. Our NLI ‘Judge’ of natural law is extremely thorough and clear. It’s slowly losing it’s ‘nicety’ and restoring its judicial clarity. It’s fascinating.


    Source date (UTC): 2025-08-06 02:46:04 UTC

    Original post: https://twitter.com/i/web/status/1952924103612219464

  • I’m just not sure yet. So far it seems like we’re replacing slaves with robots,

    I’m just not sure yet. So far it seems like we’re replacing slaves with robots, and and almost everyone will own one. As for super-ai IMO we’re facing the issue that the fundamental problem of science and technology is testing. We might make a leap (I assume we will) but will a new limit emerge? I think so.


    Source date (UTC): 2025-08-06 00:58:26 UTC

    Original post: https://twitter.com/i/web/status/1952897014959874085

  • Training an instance of GPT4

    Training an instance of GPT4


    Source date (UTC): 2025-08-06 00:55:37 UTC

    Original post: https://twitter.com/i/web/status/1952896306084802843

  • Thx. 😉

    Thx. 😉


    Source date (UTC): 2025-08-06 00:55:17 UTC

    Original post: https://twitter.com/i/web/status/1952896223398248630