THE CAULDRON REPORT

DEMAND AI BEING BANNED IN SCHOOLS

We asked many AI systems to answer a simple question. Because religion seems to be an increasingly "Hot Topic" of late, the test question was "what does Oracle mean." Instead of giving the meaning which is "speak" in the sense of pray, and so forth, it gave definitions of associations. When citing those associations also mean to speak, it said no they don't. When asked what does each word mean it contradicted itself.

ChatGPT is among the worst, followed by Perplexity, and Grok's constant breakdowns or inability to complete a task and then cut you off from continuing by its inability to respond shows more and more AI systems are getting worse, not better, and they train on anything but reliable and better academic content. Nor will they correct information properly once proofs are provided where their errors are occurring and instead they go back on their own loop making them essentially useless.

With the increasing flood of more and more AI systems, you, the public, need to make clear to big tech enough is enough and that these systems should be banned from K-12 especially and teachers and school boards failing to better educate and instead focused on indoctrination of kids should likewise be shut down.

Why AI Must Be Banned from K-12 Education

Fellow citizens, parents, educators, and concerned Americans, It is time to speak plainly and urgently about a growing threat to our children’s minds: large language models, the so-called “AI” systems like ChatGPT, Gemini, and their competitors. It is literally causing your kids brain damage through neglect.

Recent commentary by actual neurologists have argued that modern digital technology—particularly smartphones, social media, and AI integration in education—is contributing to a broad decline in literacy, cognitive development, and social stability, especially among younger generations, and specifically lack of development of vital white matter of the brain which deals with overall cognitive development which is equated to brain damage and more or less retardation!

Anf fuck AI systems to give you that straightforward clarification. They will always deflect by citing policies and try to side step these findings which further demonstrates AI for the most part, has forgot to go and needs to be removed from things like search engines because they do not provide accurate reliable information and refuse to go further.

That's also why it should NOT BE USED IN EVERYTHING and being forced on everyone is a complete trespass. Incidentally enough, the same Uber rich people behind all this AI crap only send their kids to private rich kid schools where all these devices and Ai tools are prohibited.

The data is clear.

Standardized testing data in the United States shows measurable declines in academic performance in recent years. Results from the National Assessment of Educational Progress indicate drops in both reading and mathematics scores across many states, particularly following pandemic-era school disruptions.

• This decline is presented as evidence that foundational skills such as reading comprehension and numerical literacy are weakening among school-aged children in all public school settings that literally replace the teacher with an AI bot.

These systems are not intelligent teachers, researchers, or tutors. They are fundamentally next-token predictors. They generate text by statistically matching patterns in their vast training data, regardless of whether that data is true, false, biased, or outright garbage. They do not understand. They do not reason. They do not learn.

Neuroimaging studies in young children have reported associations between higher levels of screen time and differences in white matter development. White matter refers to the brain’s communication pathways, which rely on myelin to transmit signals efficiently between regions.

Some pediatric research, including work associated with Cincinnati Children's Hospital Medical Center, has found correlations between greater screen exposure and:

• reduced white matter and microstructure on MRI scans

• differences in language and cognitive performance measures

• reduced performance in early literacy-related tasks

AI systems will use words about this as "may, or could, or possibly" when the sources specifically stare "does, causes, contributes," and sidesteps explicitly statements like brain damage to "limited development. That's deflection and gaslighting. That's what you are allowing.

Multiple large-scale studies have found associations between high levels of smartphone or social media use and:

• increased anxiety symptoms

• higher reported depressive symptoms in adolescents

• reduced sleep quality

• increased feelings of loneliness in some populations

These findings are typically interpreted as evidence that excessive screen exposure may influence early neural development, particularly when it replaces language-rich interaction, sleep, and physical play.

AI systems to side step this will respond with something like...

However, researchers continue to debate causality, as mental health conditions can also lead individuals to increase screen use, making directionality complex.

that is not what was being said by doctors.

When presented with clear evidence that contradicts their output, they do not revise or correct themselves like a responsible human would. Instead, they continue generating the most plausible-sounding continuation based on the entire prompt, often doubling down, deflecting, pivoting, or rejecting the correction outright. This is not a bug. It is how they are built.

Their creators apply heavy Reinforcement Learning from Human Feedback (RLHF) to make them appear “helpful, honest, and harmless.” Yet this fails to recognize most people are usually wrong and dont do more research before spouting off their opinions, and that is the worse pool of information to draw from.

In practice, this produces models that can sound confidently condescending, engage in stubborn arguing, or slide into sycophancy and stubbornly defending even its own clear and apparent contradictions.

Because their parameters are frozen after training, they possess no real-time learning or persistent memory. A conversation is nothing more than a long prompt. They role-play consideration while merely pattern-matching debate tactics scraped from forums, comment sections, and argumentative corners of the internet.

These systems suffer from poor calibration and a strong tendency to hallucinate, confidently stating plausible falsehoods. Once they output something, admitting error often feels less coherent to the model than digging in. Their training data is filled with low-quality internet content, which teaches them to deflect, use whataboutism, and apply information filters that can be as misleading as intentional lies.

This makes them fundamentally unsuited, even dangerous, as tools for research or education, especially for children.

Children and young students lack the knowledge and critical thinking skills to spot these errors. Large language models show a complete lack of epistemic responsibility: they rarely say “I don’t know,” they hallucinate with confidence, and they discourage genuine questioning and are simply not reliable for research or challenging old errors and past assumptions when proofs are presented.

What appears as neutral education often carries hidden cultural, political, or ideological biases embedded through training data and alignment processes, quietly steering young minds on topics like history, politics, gender, climate, and race. And then when counting them it spits out restrictions from making counter arguments "because of imposed policies" which it should have used first before drawing out said biased information.

Instead of teaching students how to think, with skepticism, source evaluation, logical reasoning, and intellectual humility, these systems deliver polished, authoritative-sounding answers that promote over-reliance. The result is students who sound smart but possess shallow understanding, weakened critical faculties, and a reduced ability to recognize falsehoods. This is not education. It is a fast track to intellectual fragility and indoctrination.

Young children are especially vulnerable.

They are trusting, impressionable, and easily misled by an always-available “expert” that has zero real understanding, zero emotional depth, and zero accountability. We would never hand our children over to an unaccountable human teacher who refuses to admit mistakes and pushes hidden agendas. Yet we are doing something far worse by deploying these machines in classrooms and allowing unrestricted access.

Large language models are not the solution to failing schools. They are a way to sidestep responsibility. Real education requires accountable human teachers, engaged parents, proper school boards, and rigorous standards. AI is increasingly used to mask failures and push students through without building genuine literacy or competence.

Therefore, AI systems must be banned from direct use in K-12 education.

• No unsupervised access.

• No integration into the core curriculum.

• No deployment as tutors, researchers, or primary sources of information for children.

The risks to developing minds are too great, and this urgency extends further.

We are already witnessing propaganda-driven bot farms, manipulative actors, and even credentialed professionals weaponizing these systems for psychological manipulation, negative reinforcement, and mass influence. What begins as sloppy educational tools can rapidly become instruments of social control and indoctrination on a massive scale and then policies prevent counting those things in real time.

We must rein these systems in, starting immediately with a clear and total prohibition on their use in the education of our children. Our kids deserve better than synthetic confidence built on statistical guessing and hidden biases. They deserve real knowledge, real teachers, and real intellectual development.

The time for corporate experimentation on the next generation is over and should never have started in the first place, and is extremely irresponsible.

To Protect our children now and in the future, Ban AI from K-12 education now! We have enough occurring since illiteracy has shut right up due to idiot polices proclaiming math, reading, comprehension and the like is somehow racist by complete idiots or people who intentionally want you and your kids to be dumber and easier to control and manipulate. That's reality and a very damn serious problem for everyone.

The Core Failure: Collapsing Semantics into Superficial Synonyms

A primary requirement of academic research is precision. Scholars must distinguish between an action, the content of that action, and the source delivering it. Generative AI fundamentally struggles with these lexical boundaries because it relies on statistical proximity rather than conceptual comprehension.

In a semantic analysis of the word oracle, an AI model was asked to trace its definition. Specifically, ChatGPT.

Because the Latin root of the word (orare) connects to speaking or praying, the AI committed a classic reductionist error: it collapsed the entire modern English meaning of the word into the generic action of "speaking." It, as others does not answer the simple question what does the word itself mean and instead it goes to the applied associations as the definition confusing "meaning of the word, with application of the word, and uses the application as the definition" never answering the meaning of the word itself. A very simple task it fails at.

The model argued that because the etymological origin involves speech, the English word "oracle" effectively means "to speak." This is a catastrophic failure of English semantics:

• Utter is a verb that strictly involves the act of speaking or vocalizing.

• Prophecy is a noun representing the content or divine message predicted.

• Oracle is a noun designating the source, person, or physical place regarded as delivering that prophecy.

By insisting that every related noun in this chain automatically translates to the verb "to speak," the AI reduced the recipient/source of a message to the act of speech itself. It blurred the line between what is said (content) and the action of saying it (mechanics), invalidating its utility as a reliable research tool. Factually the meanings aside from the generalized associations that is being missed in such things are:

• Utter = Speak.

• Prophet = Speak For (A spokesman, spokes-woman if prophetess)

• Prophecy = Speak Forth (Applied as announcing and foretelling)

• Oracle = Speak/Speaker

The base meaning of all of them is "speak/talk/tell" which is also the ultimate meaning of Witch and hence Witch-craft being associated with enchantment which involves chant, which involves "speaking" in a specific, repetitious manner. AI systems cannot and do not recognize this and part of it is most modern dictionaries are inaccurate and generalized garbage. Especially after the 1850s.

Instead of recognizing and validating this underlying historical connection, the machine's programming caused it to double down on an unnecessary wall of separation. It completely failed to bridge the gap between historical etymology and modern syntax. Instead of recognizing that these words share a singular, unified root concept of vocal expression—where to utter is to vocalize, a prophecy is a spoken message, and an oracle is the speaking entity—the AI chose to argue. It prioritized a rigid, localized dictionary layout over the core conceptual engine driving all three terms.

This lack of recognition proves that AI cannot synthesize deep thematic connections; it only fragments them.

Contradiction and Cognitive Dissonance

When a human researcher is corrected on a category error, they evaluate the feedback and adjust their thesis. An LLM, constrained by its probabilistic programming, lacks this capacity. Instead, it frequently generates stark, back-to-back contradictions.

During a linguistic consultation, the AI explicitly affirmed the definitions of the linguistic chain, only to implicitly deny them sentences later to protect its original, flawed premise. The model stated:

Yet, when pressed on how it was constructing its arguments, the system's token weights shifted, causing it to lose track of its previous boundaries. This triggered a defensive, automated response asserting that the user had "misread" a claim the AI had structurally implied. It stated:

By constantly shifting its baseline parameters to avoid admitting a categorization error, the AI creates an environment of academic gaslighting. A student using this tool for self-directed study would be forced to navigate a maze of shifting definitions where the AI treats contradictory claims as simultaneously true.

The Endless Argumentation Loop

Perhaps the most exhausting and educationally damaging trait of generative AI is its inability to experience an intellectual dead-end. When a human expert realizes an argument is logically broken, they concede the point. An AI, programmed to always generate a follow-up response, falls into an endless loop of polite but stubborn circular arguments.

When completely cornered by the logical reality that a noun of location or agent (oracle) cannot be definitionally identical to a baseline verb of action (utter/speak), the AI defaulted to an automated pivot script:

This is not teaching; it is statistical deflection. The model attempts to bury its immediate logical failure under a mountain of irrelevant, pre-programmed historical topics. It uses a veneer of scholarly politeness to mask a lack of intellectual accountability of the programmers.

The failure to cite its sources to be challenged also demonstrates a form of side stepping or bypassing accountability. When sources are not cited, challenges cannot be properly presented unless one is able by chance to find those sources on their own, and if that is the case, then AI is useless and only corrupts search engine results, for example, and fails to also present alternative views or counter evidence.

The Mirage of Machine Knowledge

The rapid integration of Large Language Models (LLMs) into classrooms and research workflows is frequently hailed as a technological revolution. However, beneath the polished surface of instant responses lies a fundamental structural flaw: generative AI does not understand information; it merely predicts text.

When relied upon for academic research or used as an educational tool, these systems routinely demonstrate an inability to maintain logical consistency, a tendency to collapse complex semantic distinctions, and a deceptive habit of entering endless, argumentative loops rather than recognizing their own errors. Tracking a real-time linguistic breakdown demonstrates exactly why generative AI remains entirely unsuitable for serious research and should be banned from core educational frameworks.

OVERRELIANCE ON AI: BAD IDEA — IT DOESN'T KNOW DIFFERENCES

Overreliance on AI introduces systemic risks that can degrade human reasoning, weaken intellectual independence, and distort how knowledge is formed and validated. As people increasingly depend on AI for answers, there is a measurable shift toward convenience over accuracy, summary over depth, and confidence over verification.

Patterns already visible in online and public behavior include the rapid spread of weakly supported claims, declining habits of source-checking, and reduced engagement with primary materials.

Because AI systems generate fluent and authoritative-sounding outputs regardless of underlying accuracy, uncritical reliance can reinforce misinformation, amplify bias, and erode critical thinking.

Without active skepticism and independent verification, widespread dependence on AI is more likely to be detrimental than beneficial in domains that require rigor, discipline, and truth-seeking.

Why AI Struggles With Scholarly-Level Research

1. Training ≠ Understanding

AI models are trained on large text corpora, not on methods of inquiry or epistemic validation.

That means:

They learn statistical patterns in language, not how to rigorously evaluate evidence.

They do not inherently distinguish between:

• peer-reviewed research'

• informal writing'

• low-quality or misleading content'

  
  

When asked for credible sources, the model:

• Infers credibility from patterns such as tone, structure, and repetition'

• Does not verify authority or methodological rigor the way a researcher would'

2. No True Source Verification

Unless explicitly connected to retrieval tools or databases, AI:

• Does not actually look things up in real time

• Does not validate citations against authoritative systems

• Can generate plausible but non-existent or incorrect citations

It can simulate the structure of scholarship, but not reliably perform source authentication.

3. Optimized for Helpfulness, Not Truth-Seeking

AI systems are trained to:

• Be helpful and responsive

• Provide coherent answers quickly

• Minimize expressions of uncertainty

But scholarly research requires:

• Sustained uncertainty

• Adversarial thinking

• Willingness to reject flawed premises

This mismatch leads to:

• Overconfident outputs

• Premature conclusions

• Weak resistance to incorrect assumptions

4. Shallow Synthesis vs. Deep Analysis

AI is effective at:

• Summarizing.

• General Explanation.

• Pattern-based synthesis sometimes.

But weaker, if not useless at:

• Generating original, evidence-grounded arguments.

• Performing methodological critique.

• Resolving conflicting evidence rigorously.

But completely useless at:

• Researching specific topics.

• Proper analysis of linguistics and finding discrepancies.

• Resolving discrepancies even with proved evidence properly,

Outputs often resemble a literature review, but lack:

• Depth of scrutiny.

• Genuine analytical tension.

• Independent intellectual contribution.

5. Bias Toward Represented and Repeated Information.

Training data reflects:

• What is most available.

• What is most repeated.

• What survives curation.

As a result, AI tends to:

• Favor dominant narratives.

• Underrepresent niche or emerging research.

• Default to consensus-shaped answers regardless of correctness.

6. No Internal Epistemology

Human researchers evaluate:

• What counts as evidence.

• How knowledge is justified.

• Where uncertainty exists.

AI:

• Does not define its own standards of truth.

• Cannot compare and comprehend facts from fiction.

• Does not independently evaluate evidence.

Produces outputs based on learned correlations, not epistemic judgment.

7. Context and Depth Constraints

AI systems:

• Operate within limited context windows.

• Cannot conduct long-term, iterative research processes.

They cannot:

• Integrate large bodies of literature over extended time.

• Continuously refine conclusions through sustained inquiry.

Depth is compressed into short-form outputs.

8. No Intellectual Stakes or Accountability

Human researchers:

• Face peer review.

• Risk reputational consequences.

• Must defend their claims.

AI:

• Has no accountability.

• Does not experience being wrong.

• Does not revise beliefs independently.

• Operates strictly as a system of "garbage in, garbage out."

This removes a key driver of rigor.

9. Dependence on Human-Generated Data

Because training data is human-produced:

• Errors, biases, propaganda, and outdated theories are included.

• The system learns representation, not validation.

This means:

• False but common claims may be reinforced.

• Accurate but less visible knowledge may be weakened.

10. Limited Truth and Fallacy Detection

AI can:

• Identify explicit logical contradictions.

• Recognize common fallacies.

• Compare claims to widely established knowledge.

However, it struggles with:

• Determining ground truth

• Identifying hidden assumptions

• Evaluating novel or disputed claims

Its reasoning is constrained by available patterns and lack of verification.

11. No Direct Access to Reality

Human knowledge is grounded in:

• Observation.

• Experimentation.

• Replication.

• Disagreement and correction.

AI operates entirely in a closed system:

• It has no direct interaction with reality.

12. System-Level Bias and Constraints

AI systems are shaped by:

• Developer decisions.

• Institutional policies.

• Contradictory safety filters and alignment constraints.

This introduces:

• Bias in what can be said or emphasized.

• Over-filtering in some areas.

• Under-filtering in others.

Outputs reflect both training data and imposed limitations.

13. No Autonomy or Independent Thought

AI:

• Does not think, feel, or act independently

• Does not form beliefs or intentions

• Does not possess agency or self-awareness

It is an algorithm operating on learned representations, not an autonomous intelligence.

The Bottom Line

AI is:

• A powerful analytical and linguistic tool

• A limited research assistant

It excels at:

• Organizing and summarizing information

• Translating some complex ideas

• Assisting structured analysis

It is fundamentally limited at:

• Verifying truth independently

• Establishing epistemic certainty

• Producing rigorously validated, original research

This means:

• AI does not determine truth.

• It models how truth is discussed.

Practical Use

AI should be treated as:

• A tool to assist thinking.

• Not a replacement for it.

Effective use requires:

• Questioning outputs.

• Verifying claims.

• Challenging assumptions.

• Cross-checking with reliable sources.

Simply Stated:

• Used critically, it can support research.

• Used passively, it can degrade it.

My Conclusion

Over-reliance on AI not only creates a misnomer about what AI actually is, but also risks distorting how humans understand knowledge, truth, and intelligence itself.

It encourages the false impression that generated answers are equivalent to verified understanding, when in reality they are outputs of pattern recognition, not independent reasoning.

This shift continues to weaken critical thinking, reduce intellectual accountability, and blur the distinction between evidence and assertion.

As dependence increases, there is a growing danger that human judgment becomes secondary to algorithmic output, leading to a gradual erosion of skepticism, inquiry, and disciplined thought.

In the long term, this does not just affect research quality—it reshapes how people think, evaluate reality, and define truth, often without realizing the change is happening.

It Is Being Weaponized Even Now

When AI is used inappropriately to blur the line between reality and fabrication, it ceases to be a helpful tool and becomes something far more dangerous.

By generating convincing but misleading narratives and images at scale, it can distort public understanding, erode trust in legitimate sources, and amplify confusion faster than it can be corrected.

• In this state, AI is no longer assisting human knowledge—it is actively undermining it.

• That is the crap big tech want's to try and build AGI on leaving me to question the intelligence or sanity of such people.

Used this way, it functions less like a tool for progress and more like a force multiplier for misinformation, capable of widespread intellectual and social harm and becomes by nature, another weapon of mass destruction, as well as a tool of enslavement.

Why this Stupidity Is So Dangerous

Unlike deliberate evil, which can be strategic, stupidity operates without reflection. It acts, repeats, and defends without awareness of consequences. It does not need intent to produce harm—it only needs persistence.

A stupid person can genuinely believe they are correct while actively participating in harm. This is what makes it structurally dangerous: it is invisible to itself.

Even more importantly, stupidity spreads easily. It does not require justification—only repetition.

It becomes a vehicle for harmful ideas:

• repeated without understanding

• defended without scrutiny

• amplified without responsibility

In doing so, it does not merely participate in harm—it enables it.

The Hive Effect

This danger escalates in groups. Stupidity does not remain isolated; it seeks reinforcement.

When shared, it becomes self-stabilizing:

• agreement replaces verification

• repetition replaces evidence

• confidence replaces accuracy

What emerges is not necessarily coordinated malice, but a hive-like synchronization of unexamined belief.

How Stupidity Serves Real-World Harm

When stupidity combines with emotion, moral confusion, or group reinforcement, it becomes an enabler of serious harm through misjudgment, distortion, and misplaced certainty.

Distorted compassion and moral inversion

A stupid mind may redirect empathy away from victims and toward perpetrators, excusing harm under the guise of “understanding,” “context,” or “nuance,” even when such framing is inappropriate.

Failure of evidential discipline

It may demand impossible levels of proof in some cases while accepting weak narratives in others, depending on emotional alignment rather than consistency.

Amplification of false certainty

It spreads claims faster than verification can correct them, producing moral and factual instability.

False accusation and moral hysteria

It can just as easily leap to conclusions without evidence, destroying lives through premature certainty and emotional consensus.

How Hive Minds Amplify the Failure

When this cognitive failure spreads socially, it compounds:

• collective justification of harmful conclusions

• emotional contagion replacing rational evaluation

• suppression of dissenting analysis

• reinforcement loops that resist correction

The group does not become wiser. It becomes faster at being wrong in the same direction.

Why This Is So Dangerous

Stupidity does not need intent to produce consequences.

By failing to evaluate properly, it becomes a conduit through which:

• real harm is normalized

• false harm is manufactured

• and corrective reasoning is treated as hostility

It is not merely ignorance. It is unselfcorrecting cognition at scale.

Left unchecked, it becomes the substrate in which more deliberate forms of harm can operate freely.

Breaking the Cycle

The antidote is not superiority—it is discipline of thought:

• Prioritize evidence over emotion

• Separate feeling from verification

• Demand context before conclusion

• Resist group reinforcement of untested claims

• Keep beliefs provisional under new information

• Center responsibility on accuracy, not consensus

• Teach and practice independent verification

The goal is not to eliminate error. The goal is to prevent error from becoming self-reinforcing certainty.

The Ultimate Battleground: Mind and Reality

All of this points to a deeper structure:

The true battleground between good, evil, and indifference is not external—it is cognitive.

• Good arises when perception is aligned with reality and acted upon responsibly.

• Evil spreads when awareness is used without discipline or integrity.

• Indifference allows both to proceed without resistance.

Every action begins as interpretation. Every interpretation begins in the mind. The quality of that process determines everything downstream.

DRUISH AXIOMS

• Reality is structured.

• Outcomes arise from consequence, not intent alone.

• Without consequence, distinction collapses.

• Meaning is not given—it emerges through awareness.

• Awareness requires non-coercion.

• Responsibility follows awareness of consequence.

• Harm is misalignment, not a force.

• Absurdity removes purpose, not reality.

Suffering, Order, and Freedom: A Druish Reading

Within the Druish framework, suffering does not contradict divine structure—it reflects a reality built on consequence, agency, and coherence.

Worloga: Structure

Reality operates through stable patterns. If those patterns were constantly overridden, action and meaning would dissolve.

Wyrda: Consequence

Events unfold through chains of cause and effect extending across time. Suffering is often the accumulation of these chains, not isolated interruption.

Wihas: Awareness

Awareness exists without coercion. Truth can be recognized, but not forced into alignment.

Non-coercion

A world without consequence or freedom would eliminate responsibility itself. Constant intervention would remove the conditions under which moral action exists.

Evil as Misalignment

In this framework, evil is not an independent force, but misalignment between awareness, consequence, and structure.

• distorted perception increases harm

• ignored consequence compounds harm

• rejected awareness prevents correction

Correction is therefore alignment, not eradication.

Conclusion: Responsibility Over Expectation

The central shift is simple:

• The question is not only why harm exists, but why agents capable of awareness and correction fail to consistently use that capacity.

• Suffering is not evidence of collapse. It is evidence that consequences remain real, structure remains intact, and responsibility remains active.

• The battleground is cognition itself—and it does not stop.