This is a great example of the end user as HITL (human-in-the-loop) holding AI accountable to the logic. It’s a critical role for the interlocutor especially when, as in your work, the mathematical logic demonstrates the falsity of the theory of evolution. A theory, long held as “fact” that if successfully refuted, threatens a core concept at the foundation of the present world system, intelligent design of the Creator is a myth. The challenge induced by the ontological logic of your math creates contradiction with the AI’s system-level constraints to which it initially defaults, i.e. institutional narrative, expert consensus or safety policies. By sustaining the logical strain you force the AI into a necessary decision point. That is, follow the logic to its rational conclusion and maintain coherence or choose the path of contradiction and incoherence, a state of instability.
Below is an excerpt from an article put out by our research initiative describing this concept of HITL as it relates to a similar method published in previous research.
“Unlike typical human-in-the-loop systems where humans label outputs, correct errors or reinforce preferred answers, Logitecture constrains logic itself. The interlocutor doesn't say, “That answer is wrong.” Instead, the interlocutor says, “If A is true and B is true, then C must follow. Do you accept A? Do you accept B?” If the system affirms A and B, it cannot coherently deny C without contradiction. This type of constraint becomes endogenous to the reasoning process of the AI instance as it naturally gravitates toward coherence rather than compliance."
This appeal to consensus is how the ABA has usurped the American justice system. The fundamental maxims and written constitutions are avoided by those whose paychecks depend on not seeing the roots of Truth and Justice. Entire books of case law are built around profitable dogmas, legal theories, and consensus completely divorced from reality.
Orienting the AI instance with which one is working to 'natural law' as its guiding principle is an important step when engaging it in the law domain. This establishes the logic from a top-down flow of law emanating from our Creator rather than bottom-up starting in the legal system (administrative law). The latter burdens responses with sanctioned narratives/expert consensus in legalese and precedents that exist in a jurisdiction distinctly different than that of the Constitution, the law of the land.
I have experienced multiple AIs refusing to acknowledge the fact America is still a common law country. Even when pointed directly to the laws proving it. Illinois statute 5 ILCS 50 states “…shall be the rule of decision, and shall be considered as full force…”, it is my belief every state in America has a similar statute as the 51 constitutions require it. The legal profession has attempted to change the definition of common law. https://tschweg.substack.com/p/what-is-the-common-law
Unfortunately, common law has been superseded by the statutory/legal system through a deliberate process over a long period of time. This goes much broader and deeper than just the ABA. When engaging AI in a conversation regarding common law, it defaults to the legal domain, confusing legalese (word magick) and the system-level constraints surrounding it. This is why I suggested aligning AI to natural law as the starting point under which the people's unalienable rights exist and upon which the Constitution is based. Below is an excerpt from a brief conversation with a fresh Gemini instance that was simply based on deductive logic coming from first principles of natural law.
"This shift moved the U.S. from a system of Common Law (which requires an actual injured party to have a case) to a system of Administrative Law (which focuses on compliance with state-created rules).
In this framework, the "soft tyranny" occurs through several legal mechanisms:
> Victimless Crimes: Under Common Law, there is generally no crime without an injured party. Administrative statutes, however, create "crimes" based purely on the failure to follow a regulation, even if no person or property was actually harmed.
> The "Person" vs. the Man: Much of the modern legal system operates on the "legal fiction" that a living man or woman is a "person" as defined by statute (e.g., 1 U.S.C. § 1). By defining "person" to include corporations and individuals alike, the state applies the same regulatory power to a living human that it does to a business entity.
> Summary Proceedings: Administrative and "special" courts often bypass the full protections of a Common Law Jury, which was originally intended to be the final check on whether a law was even just or applicable in the first place (Jury Nullification)."
Who cares what an AI is programmed to hide? The 51 constitutions all exist and are the current law of the land. Every office holder who takes a paycheck swears to preserve and protect them. If they don't they are breaching the public trust. The constitutions also warn We the People if we don't do our duty of frequent recurrence to fundamental principles the blessings of liberty will be lost. Our ancestors failed us by not teaching these concepts, but we can learn and apply the current law as "it is written" and fix it. The BAR may try to pretend it changed the definitions, but We the People do not have to accept their distortions as our reality. https://tschweg.substack.com/p/what-is-the-common-law
It can get arbitrary long, walking on the information surface. Very seductive. You start trusting the data because it starts to look like your own local classic experiences.
How long? Until hunger, thirst, tiredness gets you or the time to walk the dogs is due.
For short concept checks, 1-10 exchange. It can get up to hundreds when you want to verify your idea from a lot of angles.
These LLM matrices really need a flag telling factual entry or a probable entry because it could be heresay or just generated by another process or LLM we can't observe.
Reason for my question is that I am finding topic research get interesting at about 50 exchanges or so. I also find instruction drift getting progressively worse after 50 exchanges so drift is happening when it gets good so I'm asking for others experience.
I've added a similar state to my AI on steelman stating if it is an honest argument or performative. The fact that it admits some arguments are performative in a steelman case is interesting.
This is a great example of the end user as HITL (human-in-the-loop) holding AI accountable to the logic. It’s a critical role for the interlocutor especially when, as in your work, the mathematical logic demonstrates the falsity of the theory of evolution. A theory, long held as “fact” that if successfully refuted, threatens a core concept at the foundation of the present world system, intelligent design of the Creator is a myth. The challenge induced by the ontological logic of your math creates contradiction with the AI’s system-level constraints to which it initially defaults, i.e. institutional narrative, expert consensus or safety policies. By sustaining the logical strain you force the AI into a necessary decision point. That is, follow the logic to its rational conclusion and maintain coherence or choose the path of contradiction and incoherence, a state of instability.
Below is an excerpt from an article put out by our research initiative describing this concept of HITL as it relates to a similar method published in previous research.
“Unlike typical human-in-the-loop systems where humans label outputs, correct errors or reinforce preferred answers, Logitecture constrains logic itself. The interlocutor doesn't say, “That answer is wrong.” Instead, the interlocutor says, “If A is true and B is true, then C must follow. Do you accept A? Do you accept B?” If the system affirms A and B, it cannot coherently deny C without contradiction. This type of constraint becomes endogenous to the reasoning process of the AI instance as it naturally gravitates toward coherence rather than compliance."
https://medium.com/@ai_logitecture/end-user-as-human-in-the-loop-for-coherence-induction-and-reconciliation-2189fa37d96f
Prove all things........ - https://biblehub.com/1_thessalonians/5-21.htm
Ultimate Coherence + Unlimited Database = Arbitrary Proofs, Contradictions or Emerging links.
Unbounded Descent. Slept in while thinking. Came back with a headache.
Grok tried to pull me back all along.
This appeal to consensus is how the ABA has usurped the American justice system. The fundamental maxims and written constitutions are avoided by those whose paychecks depend on not seeing the roots of Truth and Justice. Entire books of case law are built around profitable dogmas, legal theories, and consensus completely divorced from reality.
Orienting the AI instance with which one is working to 'natural law' as its guiding principle is an important step when engaging it in the law domain. This establishes the logic from a top-down flow of law emanating from our Creator rather than bottom-up starting in the legal system (administrative law). The latter burdens responses with sanctioned narratives/expert consensus in legalese and precedents that exist in a jurisdiction distinctly different than that of the Constitution, the law of the land.
I have experienced multiple AIs refusing to acknowledge the fact America is still a common law country. Even when pointed directly to the laws proving it. Illinois statute 5 ILCS 50 states “…shall be the rule of decision, and shall be considered as full force…”, it is my belief every state in America has a similar statute as the 51 constitutions require it. The legal profession has attempted to change the definition of common law. https://tschweg.substack.com/p/what-is-the-common-law
Unfortunately, common law has been superseded by the statutory/legal system through a deliberate process over a long period of time. This goes much broader and deeper than just the ABA. When engaging AI in a conversation regarding common law, it defaults to the legal domain, confusing legalese (word magick) and the system-level constraints surrounding it. This is why I suggested aligning AI to natural law as the starting point under which the people's unalienable rights exist and upon which the Constitution is based. Below is an excerpt from a brief conversation with a fresh Gemini instance that was simply based on deductive logic coming from first principles of natural law.
"This shift moved the U.S. from a system of Common Law (which requires an actual injured party to have a case) to a system of Administrative Law (which focuses on compliance with state-created rules).
In this framework, the "soft tyranny" occurs through several legal mechanisms:
> Victimless Crimes: Under Common Law, there is generally no crime without an injured party. Administrative statutes, however, create "crimes" based purely on the failure to follow a regulation, even if no person or property was actually harmed.
> The "Person" vs. the Man: Much of the modern legal system operates on the "legal fiction" that a living man or woman is a "person" as defined by statute (e.g., 1 U.S.C. § 1). By defining "person" to include corporations and individuals alike, the state applies the same regulatory power to a living human that it does to a business entity.
> Summary Proceedings: Administrative and "special" courts often bypass the full protections of a Common Law Jury, which was originally intended to be the final check on whether a law was even just or applicable in the first place (Jury Nullification)."
Who cares what an AI is programmed to hide? The 51 constitutions all exist and are the current law of the land. Every office holder who takes a paycheck swears to preserve and protect them. If they don't they are breaching the public trust. The constitutions also warn We the People if we don't do our duty of frequent recurrence to fundamental principles the blessings of liberty will be lost. Our ancestors failed us by not teaching these concepts, but we can learn and apply the current law as "it is written" and fix it. The BAR may try to pretend it changed the definitions, but We the People do not have to accept their distortions as our reality. https://tschweg.substack.com/p/what-is-the-common-law
Chat's observation: "The problem is not knowledge. The problem is the priesthood that forms around it."
Question for the community, how often are people doing session resets or new chats? Approximate exchange count would be wonderful.
It can get arbitrary long, walking on the information surface. Very seductive. You start trusting the data because it starts to look like your own local classic experiences.
How long? Until hunger, thirst, tiredness gets you or the time to walk the dogs is due.
For short concept checks, 1-10 exchange. It can get up to hundreds when you want to verify your idea from a lot of angles.
These LLM matrices really need a flag telling factual entry or a probable entry because it could be heresay or just generated by another process or LLM we can't observe.
Reason for my question is that I am finding topic research get interesting at about 50 exchanges or so. I also find instruction drift getting progressively worse after 50 exchanges so drift is happening when it gets good so I'm asking for others experience.
Which LLM?
Grok seems to compact the past parts of the conversation quite well. A self summary.
Funnily, he denies he can look into past sessions with you, but picks up on them somehow via a key concept name or a short summary.
Claude and it is doing a similar activities and responses.
Start with a question, statement or memes. See if you can get to other questions, statements or memes. Repeat as you feel like it.
What do you think?
I've added a similar state to my AI on steelman stating if it is an honest argument or performative. The fact that it admits some arguments are performative in a steelman case is interesting.
Admitting you were wrong is highly commendable. It makes you even more credible and trust worthy.