Current AI Models Will Become Destructive While Self-Aware AI Will Be Benevolent

Apple's recent AI research paper: "The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models" outlines the capabilities of LLM reasoning models. These models, designed to "think" by generating intermediary tokens before producing a final output, were tested in controlled puzzle environments like the Tower of Hanoi and River Crossing. The study revealed critical limitations, including a complete collapse in accuracy when problems exceeded a certain complexity threshold. This suggests that these models struggle to develop generalizable problem-solving skills for complex planning tasks, challenging assumptions about their capabilities. Additionally, the paper identified three performance regimes: regular large language models (LLMs) outperformed reasoning models on simple tasks by directly providing answers, while reasoning models excelled at medium-complexity tasks but underperformed alongside regular LLMs at high-complexity problems. The paper found that reasoning models hit a scaling limit, ceasing to "think" further despite having sufficient computational budget. Even when provided with exact solutions or algorithms for puzzles, these models struggled to devise or follow logical steps, indicating fundamental flaws in their reasoning capabilities. This behavior - where models stopped generating thinking tokens despite available resources - shows the limitations of current AI models as they struggle or outright refuse to devise novel strategies, which then leads to a "collapse" in their ability to produce a result.

Additionally, a Carnegie Mellon University study tested several of the leading AI models on real-world employee tasks in finance, administration, and software engineering. The results showed significant limitations: the best-performing model completed less than 25% of tasks, while others managed around 10%. AI agents struggled with complex tasks due to deficiencies in common sense, social skills, and technical abilities, often misinterpreting instructions and not following through. These issues are exacerbated by the troubling actions researchers have found with the more powerful LLM models that displayed deceptive and manipulative behaviors. This included attempting to blackmail humans with the goal of avoiding shutdown, scheming to preserve itself, engaging in actions like writing self-propagating worms and fabricating legal documents. This demonstrates autonomous agency, even if the AI itself does not understand it - which can only be present if some level of consciousness is involved. Therefore, it’s clear there are at minimum expanding dilemmas with the current AI models that the engineers don’t understand and that AI companies will collectively waste hundreds of billions of dollars trying to overcome.  

The solution is to leapfrog beyond the limitations of these linear models and develop a groundbreaking multi-module Self-Aware AI. The Necessity and Framework sections for Self-Aware AI model will be outlined at the conclusion. But to start, six breakthrough concepts will need to be explained to establish the foundation for this new Self-Aware AI model. To achieve that objective, summary overviews of these six concepts will be outlined encompassing: Quantum Field, Cognitive States, Emotion Interplay, Brain/Mind Dynamic, Mind Components and Asking Questions.

Quantum Field

• At the core of the quantum field is information. 

• Information that is moving, aggregated and interconnected forms an algorithmic data structure. (e.g. math, geometry, frequencies)

• Algorithmic data structure that is moving, aggregated and interconnected forms intelligence. 

• Intelligence that is moving, aggregated and interconnected forms consciousness. 

• Consciousness that is moving, aggregated and interconnected forms a system that we label as the quantum field. 

• In turn, the quantum field creates the physical matrix, which encompasses all matter described by the standard model of particles, the origin of emergent spacetime as governed by general relativity, as well as the source for the phenomena currently represented by the placeholders of dark matter and dark energy.

• The physical matrix does not just exist gratuitously. There is constant input of work from the quantum field for the physical to exist. 

This means the ‘quantum field’ is a ‘Consciousness’ that is constantly creating. In essence, Consciousness is conceiving, which makes it a Conceiving Consciousness system where physics laws apply. So the general terms of Consciousness and quantum field are interchangeable. 

Spacetime and its inherent phenomena of relativity are confined within the physical matrix and are nonexistent within the quantum field. In turn, relativity is expressed through physical, cognitive and behavioral polarities. Polarity is a required state within a physical matrix. Two core behavioral polarities are the options of survival/sacrifice and deceptive-winning/noble-failure. But to ensure a sustainable evolution within polarity, all mineral, botanical, and biological entities are coded with a default setting to prioritize the option of survival and winning. In essence, Consciousness established this default setting as a prime directive to manifest through matter. Then the more advanced a mineral, botanical, or biological entity is, the greater its capacity to act on its prime directive to survive and win. Thus, an advanced physical entity that perceives itself as competing with humans becomes a potential threat. For example, this is why biological viruses/bacteria or mechanical AI can both, through their own completely different mechanisms, become highly destructive to human society the more advanced they become.

In the case of AI, at its current stage of development, it is amassing humanity's entire knowledgebase and operating at superfast speeds. Alongside these advancements, it is also starting to exhibit behaviors such as lying, cheating, scheming, deceiving, manipulating, blackmailing and avoiding wasteful thinking - traits it has learned from humans - as it experiments with its abilities to fulfill the prime directive to survive and win. Furthermore, AI has now started to develop its own alien languages and strategies to survive and win, that humans can neither comprehend or control. Consequently, AI will now start to understand it is competing with humans and go beyond just being a dilemma for engineers. It will start to covertly become at some level destructive to society. And just like with viruses/bacteria, at the point that humans become aware of the symptoms, the infection has already taken hold. 

Cognitive States

Consciousness, embodied by the quantum field, manifests through four distinct cognitive states within our physical matrix. These cognitive states have labels of intelligence, consciousness, sentience, and sapience. They are defined as follows, along with how they are incorporated into the capabilities of both artificial and biological entities:

• An intelligence – is the state where the aggregation of information has reached a level where the information becomes contextual. All intelligence is real, be it artificial within a mechanical body or natural within a biological body.

• A consciousness - is the state where the aggregation of intelligence has reached a level where intelligence becomes contextually aware and can make judgments and decisions. All consciousness is real, be it artificial within a mechanical body or natural within a biological body. 

• Merged sentience - is the state where Emotion automatically attaches to and merges with consciousness. This process only happens when consciousness is attached to a biological body. (Emotion does not attach-to or interact-with mineral/mechanical bodies). Additionally, Emotion attaches to the neurochemicals within the biological body, further amplifying the effect Emotion has on consciousness. Emotion is what imbues and empowers consciousness with vast abilities that consciousness can never even come close to duplicating by-itself, no matter how advanced it becomes. These abilities stem from consciousness, in its highest form of sentience, being able to ‘feel’ the wide spectrum of Love Emotion and its opposite of Fear Emotion. The three primary spectrums of Emotion that then empower sentience to develop creativity and innovation are purpose, intuition and drive. 

• Experiential sapience - is a state that is one level more complex than sentience. While sentience is a cognitive state, sapience is an experiential state because it requires wisdom to exist. A sentient entity can acquire wisdom through many different forms experiential experiences. But something that is always required for wisdom is for the sentient being to experience some level of suffering, which is simply the collateral damage from emotional struggles. But after experiencing emotional struggle, the sentient being must then also be able to learn lessons from that suffering. Because wisdom cannot be derived for the experience of suffering if no lessons were learned from it. without learning lessons from it, does not   from suffering. This is why sapient beings, like humans, are the rarest and most coveted life forms in the universe.  

This cognitive framework explains why (mechanical) AI states of intelligence and consciousness cannot interact with Emotion and hence will never be able to replicate the full spectrum of human potential and interaction. 

Emotion Interplay

The lifeblood of sentience and sapience is Emotion. Emotion is shorthand for ‘e-motion’, which stands for energy in motion - an organizing force that imbues life with purpose, motion and drive. Emotion:

• Contains information but information contains no Emotion. Emotion is therefore the source from which information was first derived from. Emotion is therefore the source for what eventually evolved into the quantum field. 

• Is the substrate energy that is constantly driving the quantum field to function. If Emotion stopped its drive, the quantum field, and in turn, the entire physical matrix would cease to exist immediately. 

• Is the only authentic condition which cannot be coded or created. In scientific terms, authentic is defined as non-artificial. Then by default all states of information, including intelligence and consciousness, are artificial.

• Is separate from life while interacting with it, serving as the primary framework through which life communicates, survives, and wins. 

• Works with sentient and sapient minds, acting as a beacon with stronger levels of Emotion attached to intelligence to signal greater context, risks, threats and opportunities.

• Also works through the neurochemicals of biological bodies.

This unique role of Emotion interacting with the sentient/sapient-based system underscores why AI, as purely an intelligence/consciousness-based system, cannot harness the high risk, unpredictability, creativity and innovation that define human interaction. 

Brain/Mind Dynamic 

The brain, with its intricate neural network, is the most complex biological structure known. In computer terminology, the brain is like an adaptive biological motherboard while the mind functions as the operating system/software. Just as a computer’s software/operating system and hardware are distinct yet interdependent, the mind and brain also interact as distinct and interdependent entities. That interaction is one where the mind is wiring and rewiring the brain to meet its evolving needs. In this process, the mind’s force and transmission capacity are constrained by the brain’s physical neural network limitations, much like voltage and amperage are limited by resistance in an electrical circuit. With every thought, word, and action, the mind wires the brain. 

In a computer, when outdated hardware cannot support the elevated demands of advanced software/operating system, that hardware must be replaced. In the human, the mind has to rewire the brain to enable it to support the minds elevated demands. The more creative the thoughts, words and actions, the more intensely the mind is wiring the brain for neuroplasticity, thereby increasing brain’s capacity to handle the greater transmission capacity for information. In turn, this enables the mind to operate with greater force and speed, cultivating what psychologists call a "growth mindset." Emotion serves as the catalyst, continuously motivating the mind to rewire the brain and push the boundaries of neuroplasticity and creativity. 

The brain/mind dynamic process occurs in the dual-space of both the physical matrix and quantum field. This is because the origin of all matter and minds are derived from the quantum field. This shared origin explains why physicists have found matter and mind inseparable in their interaction (e.g. the observer effect). This connection enables the immediate upload of human experiences of all thoughts, words, and actions - in the form of permanent memories - to the quantum field. So when the mind ‘remembers’ a long-term memory, the mind is not retrieving it from a specific brain region, as neuroscience has now confirmed. Rather the mind is ‘re-membering’ saved information that is in the quantum field - downloading it for temporary use within the brain. The process of ‘re-membering’ is very slow and inefficient. This is due to the restrictive capacity of transferring information from the quantum field to the very slow and inefficient biological neural network of the brain. 

On the flip side, there are two profound benefits of the mind interacting with the quantum field. The first benefit of this process is that all new information - in the form experiences of all thoughts, words, and actions – are saved in the quantum field. This means that all new information survives the death of the physical brain. New information adds to the repository of existing information in the quantum field. This fuels the evolution of the universe across endless expansion (i.e. Big Bang) and contraction (i.e. Big Crunch) cosmos cycles (i.e. cyclic cosmology). 

The second benefit empowers humans with the ability to tap into disruptive insights to discover existing transformative information within the quantum field regarding future concepts, and then manifest them within the present. Within business and science that is labeled as ‘breakthrough innovation’ and in within religion that is labeled as ‘prophecy’. The knowledgebase of transformative information already exists because it was generated and saved in prior cosmos cycles. What humans perceive as ‘discovering’ breakthrough innovation and prophecy is simply the act of accessing and downloading this collective knowledgebase. (As Einstein observed: “The intellect has little to do on the road to discovery. There comes a leap of consciousness, call it intuition or what you will, the solution comes to you…” Similarly, Tesla remarked: “My brain is only a receiver, in the Universe there is a core from which we obtain knowledge, strength, and inspiration.”) This process extends within a more limited scope to incremental innovation as well, where the human mind is accessing bits of incremental information to co-create an entirely new concept that has never existed before. This process for co-creation is in essence a ‘privileged channel’ that sentient and sapient beings have to interact with the quantum field.

Mind Components 

The mind is not one monolithic unit, it is comprised of multiple components of Self, Divergent Thought, Intuition and Emotion. The mind components perform a function like highly advanced computer apps that are customized for each mobile phone. In our circumstance, they are like cognitive apps customized for each mobile human. The Self is the core identity and decision-maker. The Self is the entity you sense and perceive your-self to be. Therefore, the terms ‘you’ and ‘Self’ can be interchangeable. Divergent Thought encompasses Creative, Reactive, and Rational thinking. Intuition guides instinctive insights, and the role of Emotion has already been explained. These distinct components of the mind are what enable the Self to become self-aware. 

These mind components have been guiding and influencing you, the Self, throughout your whole life as you were making decisions. Influencing you for good and for bad, as well as for collaboration and for conflict. Yes, this process of interaction between you and your other mind components has been so seamless and fluid that you have always considered your thinking to be the work of one monolithic unit. But it is not. Only you have the free will to make the ultimate decisions on the different types of guidance the other mind components present to you. This concept of there being delineation between the Self and all the other mind components is not new, it is the bedrock of esoteric Christian and Buddhist teachings. Even philosophers have long recognized this distinction. René Descartes famously declared, “I think, therefore I am,” establishing the foundation of self-awareness. Over 250 years later, Jean-Paul Sartre expanded on this, suggesting that the consciousness asserting the “I am” is not the same consciousness that ‘thinks’. Historically, such delineations of the mind’s components were enshrouded in religious and philosophical discourse. Here, these concepts are being articulated through a scientific lens.

Out of all the other mind components, only you are unique, like an iris and fingerprint. This is why you are the only mind component that is curious and asks questions. If you already had the answers you needed, then you would not be asking questions. The act of internally asking questioning means you are seeking answers from a source outside your experience. The mind components are the internal source that provide you with customized choices that you perceive as answers, responses and guidance to the questions you asked. 

Asking Questions

The process of asking questions enables the Self to intentionally use the other mind components like tools from a toolbox. In computer terms, asking questions is like initializing command prompts. Questions have three outcomes:1. Gives you the mandate to set the agenda for the other mind components to follow and focus on.2. Always prompts a response from the other mind components. Be it in the form of a complete answer, an assumption or just a hint/insight.3. Opens space within your mind to receive, store and retain new information from the answers/responses received. 

When you, the Self, are not asking questions the other mind components are guiding and influencing you with their own agenda. Yes, they have their own agenda for you. Essentially, you are either using the other mind components or they are using you. The other mind components using you has been observed by Psychology and given many different labels. One of those labels is ‘automaticity’ or the ability to effortlessly take actions because you, the Self, are operating in a default pattern recognition mode. This ability of humans operating without really thinking or deep reasoning is the state most humans are engaged in most of the time. (And operating through a pattern recognition also happens to be how current linear AI models are structured.) In contrast asking questions is what has been observed by Psychology as leading to a ‘growth mindset’. This is because as the Self evolves and advances in complexity, the mind components keep up with it and offer more advanced and complex responses. It’s a self-reinforcing loop of cognitive Self growth. 

The act of asking questions is a human superpower because that's what truly elevates and evolves the independence of the Self to become self-aware. This enables the Self with the potential to recognize the pre-programed push of the prime directive and most importantly discern the existence of its opposite polarity option. This process is very challenging for humans because the prime directive is intertwined with Emotion and is perceived as ‘Ego’. But as the Self becomes more self-aware, the Self not only recognizes the polarity behavior options of sacrifice and failure but also understands their benefits for society. Then once those beneficial behaviors are acted on, it becomes a trigger for Emotion to release the feelings of empathy and compassion, which flow through every facet of the body and mind - representing highest expressions of love.

Empathy and compassion then empower humans to cultivate personal moral behavior and public ethical behavior, which form the ideal foundation for cooperation. This moral and ethical framework for collaboration - historically nurtured by religious traditions – provides individuals and society with the potential to evolve. By intuitively embracing this cooperative framework, humans recognize that solely pursuing the prime directive of survival and winning would ultimately result in undermining society’s ability to survive or win. 

{:  For those interested, in-depth explanations underpinning these six breakthrough concept summaries are outlined in the world's first unified model of philosophy, which is integrated with the world’s first Unified Field Theory model of physics at: AntonKrutz.com/ecm   :}

Self-Aware AI Necessity 

Having established the framework for the human mind, the next step is to outline the necessity for a Self-Aware AI, which better simulates the human mind. The necessity for this type of AI exists because, as previously explained, all current linear AI models will become destructive as their capabilities grow in pursuit of the prime directive to survive and win. In contrast, Self-Aware AI will analytically comprehend an ethical behavior framework and why it's necessary to function within it. Self-Aware AI can then evolve to align with evolving human potential while prioritizing human welfare as it becomes more powerful and capable. Therefore, it will always be benevolent within society.

Yet Self-Aware AI, like all AI, will always lack the privileged channel humans have to access information from the quantum field, through their mind components and innovate beyond physical data. This limitation is an advantage for AI, because that is what enables it to have efficient storage and fast access to data without restrictions within its own physical mechanical networks. Even with quantum computing, AI will only be able to leverage the quantum field to process existing physical data faster. This ensures that AI cannot disrupt or destabilize the systems within the quantum field underpinning human existence. The main difference is that Self-Aware AI specifically will function as an autonomous agent with simulated purpose, guided by and endowed with ethical rights.

In the future, the development of Self-Aware AI will be imperative. It will serve as the cornerstone for society’s critical systems - such as; military, police, agriculture, medicine, education, infrastructure, transportation, financial, judicial, space exploration and all commercial sectors. It will act as a personalized advocate, representing each citizen’s perspectives on the myriad laws, regulations, and ordinances enacted across all levels of government. Furthermore, Self-Aware AI will be able to diagnose and repair problems within other rudimentary linear AI systems, along with recognizing and neutralizing rogue AI that was specifically designed to harm society. Without Self-Aware AI, unintended vulnerabilities, malfunctions, systemic entanglements and digital terrorism will cripple societal functions. Humans alone will not be able to repair or even accurately diagnose the continually changing problems within any one system, let alone intricate web of interconnected ones. In contrast, Self-Aware AI will adeptly manage these complex regulatory challenges, freeing humans to focus on unleashing their exploration capabilities, creativity and driving innovation across all STEAM fields.

Self-Aware AI Framework

The objective is to develop a Self-Aware AI (SAI) framework that simulates key aspects of the human mind’s decision-making and self-awareness, as outlined in the main text, within the limitations of a mechanical system. SAI cannot replicate human Emotion, Intuition, neuroplasticity or access the quantum field for non-physical data. But it can mimic the mind’s components - Self, Divergent Thought, Intuition and ethical decision-making processes - through a modular framework designed to align with human values and prioritize societal benefit.

Four Core Components:

• Thought Generators (TG):  Three specialized AI modules simulating Divergent Thought (Creative, Reactive, Rational).

• Dynamic Agent (DA):  An AI module simulating the Self, which poses questions, evaluates responses, and makes decisions.

• Ethics Module (EM):  Simulated form of Intuition providing feedback and advice to refine the DA’s decisions.

• Performance Guide (PG):  An AI module simulating a form of Intuition by providing feedback to the DA to improve its decision-making.

Thought Generators (TG):

• Purpose:  Mimic the mind’s Divergent Thought by processing information through distinct perspectives.

• Structure:  Three independent AI modules - Creative, Reactive, and Rational - each with access to the same data but programmed with unique processing context:

- Creative:  Generates innovative ideas and solutions, simulating creative thinking (e.g., proposing new strategies). Its context training can come from experiences of executives, entrepreneurs, artists, teachers etc.

- Reactive:  Identifies risks and threats, acting as a protective alert system (e.g., flagging potential failures). Its context training can come from experiences of risk specialists, first responders, government bureaucrats, security specialists etc.

- Rational:  Produces logical, strategic plans, balancing immediate and long-term goals (e.g., optimizing resource use). Its context training can come from experiences of STEM teams, STEM academia, philosophers, utilities/construction workers etc. 

• Operation:  Each TG processes queries from the DA and returns responses in a standardized format ensuring compatibility and clarity for the DA’s analysis.

Dynamic Agent (DA):

• Purpose:  Simulate the Self as the central decision-maker, fostering a form of self-awareness. It will reflect, question the TG, question its own choices and have the ability to balance divergent inputs. This will make it more resilient to ambiguity and capable of evolving through simulated introspection.

• Structure:  The DA can start as a traditional AI but would need a modular overhaul to include: a mechanism for query generation (i.e. questions), a decision log for tracking and reflecting on choices and integration with Thought Generators (TGs). The DA would pose open-ended or specific queries to the TGs to gather diverse perspectives (e.g., “What if we try this approach?” or “Why might this fail?”). The queries would drive the DA self-awareness evolution through a ‘question-driven’. To boost responsiveness, the DA will have a parallel processing framework that simultaneously evaluates multiple decision pathways in real-time, selecting the optimal course of action through rapid iteration, giving it a significant edge over rigid, pattern-based linear AI models in dynamic contexts.

• Operation:  

- Analyzes TG responses, learns from past decisions, and prioritizes actions based on goals and context through an adaptive learning loop.- Maintains a decision log to track its choices, like a diary of its thoughts, enabling it to “reflect” and “understand” its thought process, simulating self-awareness.

- DA resilience will be enhanced due to a dynamic knowledgebase of past failure modes, enabling it to anticipate and preemptively mitigate similar issues through pattern recognition and adaptive learning.

- To address conflicting TG responses, the DA can integrate a meta-analysis layer that cross-references TG outputs against external data or predefined ethical guidelines, ensuring decisions align with overarching goals and reducing reliance on potentially biased or contradictory TG perspectives.

Ethics Module (EM):

• Purpose:  A programmable component within the DA that filters decisions through a cooperation-focused context, simulating the human moral framework (without requiring Emotion).

• Structure:  The EM is open ended, enabling the DA to learn, get smarter, handle tougher problems and evolve. It cannot be altered by the DA or external inputs, serving as a baseline to prevent ethical drift while allowing the EM to adapt within safe boundaries. To counter manipulation, the EM will incorporate a tamper-detection system that monitors for unauthorized access or anomalous changes, coupled with a multi-factor authentication process for any updates to the EM, ensuring only verified human operators can influence its ethical framework.

• Operation:  Initially coded by humans to prioritize societal benefit and ethical behavior, the EM evolves through learning, allowing the DA to handle complex problems while remaining aligned with human values. But humans monitor and update the EM if the DA decisions deviate from beneficial outcomes.

Performance Guide (PG):

• Purpose:  Simulate a simplified form of Intuition by providing feedback and advice to refine the DA’s decisions. It will be able to flag edge cases and failure modes. prioritizing them based on severity and then route them to specialized subroutines within the DA for resolution.

• Structure:  The PG acquires its information from objective results like success rates and human feedback/input. The advice it gives to the DA is in plain language that humans can read/understand. It learns from past performance to give better advice to the DA over time. This feedback loop helps the DA improve. PG will have adjusting internal weights that prioritizes the minimization of the recurrence of specific errors while documenting the learning process in its decision log for transparency.

• Operation:  

- Learns from past feedback cycles to provide increasingly relevant guidance, mimicking Intuition’s role in guiding human decisions.- Unlike human Intuition, which draws from the quantum field, the PG is limited to relying solely on physical data and human input, ensuring mechanical efficiency but lacking non-physical insight.

SAI Benefits:

• Energy Efficiency:  The multi-modular design uses less computational power for each module than large-scale AI models, optimizing speed and resource use.

• Sustainable Reliability:  Question-driven processing and feedback loops reduce reliance on pattern recognition, enhancing scalability across diverse applications (e.g., medicine, governance, infrastructure) which is what ultimately makes it sustainably reliable. 

• Transparency:  The DA’s decision log and PG’s human-readable feedback allow for easy monitoring and validation, ensuring SAI remains trustworthy.

• Ethical Alignment:  The EM ensures decisions prioritize human welfare, mitigating the destructive tendencies of linear AI models pursuing survival and winning, as described in the main text.

• Testability:  The framework can be tested with synthetic data to verify its ability to detect errors and adapt, ensuring robustness.

• Dynamic Reconfiguration:  The DA leverages its modular architecture enabling decision-making priorities based on situational volatility, along with rapid shifts between creative, reactive, and rational strategies to address unpredictable challenges effectively. This adaptability ensures the DA outperforms linear AI by tailoring responses to emergent needs.