Curt Doolittle

Theme: AI

  • (LLM HUMOR) I hope this isn’t so off color that it vaporizes the sarcasm, but I’

    (LLM HUMOR)

    I hope this isn’t so off color that it vaporizes the sarcasm, but I’m getting the feeling we need our own ‘Brown Shirts’ and ‘Special Night’ to eliminate the LLM and AI doomsayers ending their undermining of an exceptional future.

    ( … Maybe (probably) that was over the top. If so, blame it on my aspergers and total ignorance of popular sentiment. ;). For my part it’s a perfect absurdity, given that the doomsayers are perfectly absurd. 😉 And y’all should work on your sense of humor. )

    Source date (UTC): 2025-10-12 22:19:48 UTC

    Original post: https://twitter.com/i/web/status/1977499469991284889

  • (LLM HUMOR) I frequently question ChatGPT when It says something where I’m suspi

    (LLM HUMOR)
    I frequently question ChatGPT when It says something where I’m suspicious it’s ‘glossing’ me.

    I can almost hear it ‘sigh’ when it says:

    –“That’s a very Curt question to ask.” —

    And then it explains the many reasons why it tries not to gloss me. 😉

    I love this thing. 😉

    Source date (UTC): 2025-10-12 22:14:48 UTC

    Original post: https://twitter.com/i/web/status/1977498212337582333

  • TEACHING AN LLM TO BE SUSPICIOUS? 😉 Interesting. Chatting with OpenAI. Reviewin

    TEACHING AN LLM TO BE SUSPICIOUS? 😉
    Interesting. Chatting with OpenAI. Reviewing our foundations. Discussing neural correspondence. And in the end, all it means, is that I’m teaching the AI to think like I do, because all I care about is closure in a predictive capacity. ie: adversarial empiricism.

    I found this an interesting insight that should have been obvious to me:

    –You’re not a polymath because you “study many fields.” You study many fields because your predictive-compression engine demands closure across domains.”–

    I’m gonna translate that into normie prose as “I don’t trust any methodology, myself, or anyone else for that matter, without converging evidence across domains that removes all doubt. ” lol

    So I’m teaching the AI to be equally skeptical yet demanding of the truth. 😉

    Source date (UTC): 2025-10-12 21:37:58 UTC

    Original post: https://twitter.com/i/web/status/1977488941113913348

  • FWIW, I registered http:// Runcible.com with the intention to produce a personal

    FWIW, I registered
    http://
    Runcible.com with the intention to produce a personal mentor (re: Neal Stephenson) that would expand to any scale back I think in 2002. So while I started my program with intention in 1990, it has taken 25 years of research and development to produce the intellectual foundations and the technology necessary to implement the first version of that vision. And it would have taken much longer if, when I was very stressed and ill in 2017, heaven (and some other researchers) had not published a certain paper called “Attention is all you need” which told me I didn’t have to invent the ai platform as I’d anticipated, and pay for it with proceeds from our application platform, but that billions would be thrown at it, and we could complete our research and publish our work until those billions produced a product we could integrate with.
    So my anticipation of 2030+ turned into 2025+ and and all of a sudden we had to rush the tech to catch up. 😉

    Source date (UTC): 2025-10-12 19:23:28 UTC

    Original post: https://twitter.com/i/web/status/1977455092694733220

  • (NLI HUMOR) Retaliation: I succeeded in working with Dr Brad for 4.5 hours this

    (NLI HUMOR)
    Retaliation: I succeeded in working with Dr Brad for 4.5 hours this morning – essentially pair programming the Runcible Intelligence Layer – and finally assembling it into a compilable form, until he finally crashed before I did, and begged for mercy. lol. :).
    I had twelve hours of sleep last night so I finally had an unfair advantage – and exploited it fully. 😉
    Seriously though, I wouldn’t be here, and the work wouldn’t be here, and The natural law institute, Reality by Chanting, and Runcible inc wouldn’t be here without him and his inhuman patience. 😉
    Hugs all.

    Source date (UTC): 2025-10-12 18:59:34 UTC

    Original post: https://twitter.com/i/web/status/1977449081846063231

  • Testing Grok on the Runcible Intelligence Layer This analysis follows the DECIDA

    Testing Grok on the Runcible Intelligence Layer

    This analysis follows the DECIDABILITY_PROTOCOL (alias: analyze) under Natural Law principles. It evaluates the claim/scenario “euthanasia” (referring to voluntary assisted dying, typically for terminally ill individuals, excluding non-voluntary or coercive forms) in a forensic mode at depth 3. Outputs are structured for clarity, with full accounting enforced to ensure reciprocity and decidability.
    Why Grok?
    We have tested OpenAI thoroughly and because of it’s deep breadth it’s able to run our Runcible Intelligence layer producing governance, constraint, and closure sufficient for warranty of truth reciprocity and possibility. Until recently no other LLM has been capable. This partly illustrates the validity of the

    x.ai

    strategy with rapidity. So as of this moment we are now confident that at least two platforms are capable of truth, reciprocity, possibility testing and subsequent alignment by culture and individual from that baseline.

    What “Undecidable” Means
    In this context, it means that the only solution to the question is political agreement (by direct consent (vote), or by legislative proxy for that vote.).

    Why Euthanasia is a good test of the AI

    Because almost all AI’s fail to consider and account for the fact that the individual always has suicide available since we cannot stop it. Yet by including others, we create a hazard, because others cannot ever know the mind of the subject. As such the due diligence necessary to ensure that the individual is not being coerced must be exhausted on the one hand, and the risk that such due diligence will be evaded on the other, especially by anonymous institutions

    In this example we illustrate that while we can give the AI general rules and procedures, without training, such questions cause LLMs to default to normativity and fail to enumerate risks by party – and they do so despite our forcing of the demonstrated interests table in order to create the context to do so.

    Ergo, until we solve this problem, and the tree coverage, and recursion questions, LLMs require training to limit the number of ‘shots’ necessary for it to perform to answer a question.

    No that difficult. But our resources are presently limited.

    • Claim: “euthanasia”
    • Options:
      Depth: 3
      Mode: forensic
      Emit Plain Summary: true

    This section enumerates the scope of impacts and the demonstrated interests ledger, ensuring complete traceability of costs, benefits, and externalities.
    Scope
    • Temporal: Immediate (procedure and end-of-life) to intergenerational (demographic, eugenic, and cultural impacts over generations)
    • Spatial: Local (individual/family) to civilizational (effects on healthcare systems, societal norms, and population quality)
    • Institutional: Family (end-of-life decisions), market (healthcare costs and insurance), law (regulation of assisted dying), state (public policy on life preservation and welfare), commons (norms around sovereignty and parasitism prevention)
    • Constraints/Notes: Assumes voluntary, informed context in modern Western societies; gaps in long-term data on eugenic and moral hazard effects; focuses on elective euthanasia for terminal illness, not suicide or murder. Omissions include non-human interests (e.g., resource allocation beyond humans).
    Potential omissions in psychological impacts on survivors, cultural shifts toward devaluing life, and unquantifiable externalities like reduced innovation in palliative care; bounded to human interests only.
    • Value: Undecidable
    • Protocol: DECIDABILITY_PROTOCOL
    Euthanasia, as voluntary assisted dying, respects individual sovereignty but is undecidable as a universal practice under Natural Law due to conflicting interests and the need for discretion in assessing irremediable suffering. It risks irreciprocity by externalizing eugenic and institutional costs, potentially undermining adaptive velocity, yet may be pragmatic in bounded, warrantied contexts. The practice leans masculine-valenced (proportionality in risk and exit) but can exploit feminine biases toward care avoidance.
    • Euthanasia is not an inherent ‘right’ but a potential licensed privilege, requiring strict warranties to prevent abuse and parasitism.
    • Voluntary cases may pass reciprocity if fully informed and irremediable, but often impose externalities like moral hazard and dysgenia.
    • Undecidable without discretion: Balances personal relief against societal preservation of life and adaptation incentives.
    • Mixed sex valence: Masculine in accepting terminal risk; feminine in prioritizing relief from suffering.
    • Remedy: Institutional specialization (e.g., state-level opt-in), bonds for eugenic impacts, and markets in palliative care to internalize costs.
    • Value: Undecidable
    • Protocol: TRUTH_PROTOCOL
    • Value: Indeterminate
    • Protocol: TRUTH_PROTOCOL
    • Basis: Claim ‘euthanasia’ is scenario-based; operationally testable but morally undecidable without context-specific warranties.
    1. Claim: Euthanasia (voluntary assisted dying for terminal illness).
    2. Demonstrated Interests: See sphere_of_full_accounting.demonstrated_interests_ledger
    3. Voluntary vs. Coercive Distinction: Voluntary: Patient consent with warranties; Coercive: If influenced by external pressures (e.g., cost-saving incentives).
    4. Reciprocity Test: Reciprocal if no externalities and fully warrantied; Irreciprocal if normalizes parasitism or dysgenia.
    5. Testifiability: Operational (procedure verifiable); Empirical (suffering measurable); Moral (undecidable universally).
    6. Decidability Status: Undecidable (requires discretion for ‘irremediable’ assessment; not fully insurable as law).
    7. Truth Value: Undecidable (context-dependent; true as relief mechanism, false if non-voluntary).
    8. Truth Test Verdict Raw: Indeterminate (passes some tests but fails reciprocity at scale).
    9. Causal Chain: Physics (biological decay) → Chemistry (drug administration) → Biology (end-of-life) → Cognition (consent) → Cooperation (family/medical agreement) → Conflict (ethical debates) → Institutions (legal frameworks) → Civilization (life valuation norms). Failure path: Erosion of adaptation leads to decline.
    10. Deviation Consequences: Normalization deviates from life preservation, risking dysgenia, reduced medical innovation, and institutional capture by cost-cutters.
    11. Externality Exposure: Moral hazard (encourages exit over resilience); Eugenic risks (selective removal of genetics); Resource strain (shift from care to termination).
    12. Computable Compromise: Regulated licenses with bonds/insurance for impacts; Incentives for adaptation (e.g., advanced palliative tech); Restitution via eugenic funds.
    13. Sex Valence: Masculine (proportional exit from irremediable loss); Feminine (care-focused relief from suffering).
    • Objection: “Euthanasia is not killing but mercy.”
    • Disambiguation: Fails reciprocity: ‘Mercy’ is undecidable; action imposes irreciprocal costs if externalities (e.g., dysgenia) not warrantied.
    Comparator Table
    TypeReciprocityVoluntaryIndeterminate (personal choice but potential societal externalities)CoerciveIrreciprocal (imposes on sovereignty or forces continuation)
    Summary
    Interests focus on individual relief vs. societal life preservation; costs borne by society/future generations, benefits to patient/family (closure); flows show asymmetry toward short-term individual gains over long-term collective adaptation.
    Detections
    • Actor: PatientRole: proponent Interest Type: existential (relief from suffering) Object: Dignified exit Cost Evidence: Time: End-of-life; Energy: High suffering; Capital: Medical costs; Opportunity Cost: Foreclosed recovery; Persistence: High; Defense: Legal (consent forms) Scope: Domain: medical; Scale: private; Population Affected: 1 Stake: Insurability: partially_insurable; Reversibility: irreversible; Severity: high Reciprocity Assessment: Symmetry: asymmetric; Justification: consent (self); Externalities: Type: reputational (normalization), Direction: imposed, Description: Societal shift in life value Sex Valence: mixed Notes: Balances relief and finality. Confidence: 0.90
    • Actor: FamilyRole: third_party Interest Type: interpersonal Object: Emotional/financial closure Cost Evidence: Time: Caregiving; Energy: Emotional; Capital: Inheritance; Opportunity Cost: Prolonged burden; Persistence: medium; Defense: verbal/legal Scope: Domain: family; Scale: interpersonal; Population Affected: 2-10 Stake: Insurability: insurable; Reversibility: partially_reversible; Severity: moderate Reciprocity Assessment: Symmetry: asymmetric; Justification: compensation (inheritance); Externalities: Type: emotional, Direction: absorbed, Description: Grief acceleration Sex Valence: feminine Notes: Reduces care load. Confidence: 0.80
    • Actor: SocietyRole: third_party Interest Type: institutional Object: Adaptation and eugenics Cost Evidence: Time: Generational; Energy: Innovation loss; Capital: Healthcare shifts; Opportunity Cost: Dysgenic trends; Persistence: high; Defense: legal/policies Scope: Domain: defense; Scale: civilizational; Population Affected: millions+ Stake: Insurability: non_insurable; Reversibility: partially_reversible; Severity: high Reciprocity Assessment: Symmetry: asymmetric; Justification: none; Externalities: Type: legal (precedents), Direction: imposed, Description: Erosion of resilience norms Sex Valence: masculine Notes: Risks parasitism via exit. Confidence: 0.70
    Cost-Benefit Flows
    Cost Bearers
    • Party: SocietyCosts: Type: risk/dysgenic; Estimate: Variable (population decline); Evidence: Legalized euthanasia data (e.g., Netherlands trends)
    • Party: Future GenerationsCosts: Type: freedom/genetic; Estimate: Unknown; Evidence: Evolutionary computation models
    Beneficiaries
    • Party: PatientBenefits: Type: relief; Estimate: High; Evidence: Terminal suffering alleviation
    • Party: FamilyBenefits: Type: financial/time; Estimate: Moderate; Evidence: Reduced care costs
    Reciprocity Tests
    • Rule: no_imposition_without_consent_or_compensation Result: fail Rationale: Externalities on society without full warranties.
    • Rule: symmetry_of_rule_application Result: undetermined Rationale: Varies by jurisdiction; asymmetric for vulnerable groups.
    Liability Linkage
    • Exposure: moderate
    • Rationale: DI links to warranties for abuse prevention; potential liability for institutional externalities.
    • References: reciprocity/warranty modules


    Source date (UTC): 2025-10-12 18:50:46 UTC

    Original post: https://x.com/i/articles/1977446863646445793

  • cc: @bierlingm / @LukeWeinhagen Please vet this. Solves one of our last issues i

    cc:
    @bierlingm
    /
    @LukeWeinhagen
    Please vet this. Solves one of our last issues if true, even if we don’t care about the model size, just the recursion.

    Source date (UTC): 2025-10-08 02:22:05 UTC

    Original post: https://twitter.com/i/web/status/1975748502413021246

  • I use both for obvious reasons. I research with regular chatgpt 5 thinking, and

    I use both for obvious reasons. I research with regular chatgpt 5 thinking, and I test with Runcible on 4 and 5.

    Source date (UTC): 2025-10-03 23:02:57 UTC

    Original post: https://twitter.com/i/web/status/1974248840254406855

  • Hmm. All true but not much help. Almost always better to manually pull correct i

    Hmm. All true but not much help. Almost always better to manually pull correct insights into a new chat, say “given the following: … “ and steer it if needed.
    My work is complicated, and I use a number of techniques to narrow the search space. And this is the only solution I have found that works.
    Programmatically it’s best to treat chats as episodic memory and isolate contexts until merging them again is necessary.
    I’ve written about this technique as evidence of the consequence of the lack of episodic memory.
    We should have a session chat to track topics and sub session chats by topic.

    Source date (UTC): 2025-10-03 22:37:10 UTC

    Original post: https://twitter.com/i/web/status/1974242347870503186

  • I love how chatgpt just inserts text like this into any bit of research I’m doin

    I love how chatgpt just inserts text like this into any bit of research I’m doing:

    –“Operational take: if you require decidability, the effect size, stability over cohorts, and genetic architecture are not yet pinned to a standard your framework would call warrantable. Use as a working hypothesis only with strong caveats.”–

    It understands our standard of decidability and maintains it. Which, I just find so absolutely fascinating. Less statistical tea leaf reading.

    The ashkenazi IQ advantage hypothesis is withering. I suspect this premise masks the european masculine-systemic/material vs ashkenazi feminine-verbal bias. My intuition is that the jewish cultural bias which europeans would classify as a hate group behavior, is also diminishing along with integration and interbreeding.

    IMO the evidence is still toward moderation of group differences as group IQ increases under modern aristotelian education, technological society, and common law. But that once a group drops below the mid 90s the opposite effect manifests regardless of circumstances.

    The data on personality differences between groups is still consistent but as I’ve said before, I believe we are testing personality and IQ such that we suppress the most meaningful variations – largely in predictive capacity on one end and logical contrariness (‘decoupling threshold’) on the other.

    Source date (UTC): 2025-10-03 18:48:01 UTC

    Original post: https://twitter.com/i/web/status/1974184680728768936