towards an infinite
state of being
amplifying human consciousness with machines
shravan nageswaran, sumon sadhu, maxim solovyev, mehdi zadem
MARCH 2024. [20 minute read]
How can we design a new type of intelligence system that can benevolently accelerate the progress of humanity?
How can we design a new type of intelligence system that can benevolently accelerate the progress of humanity?
How can we design a new type of intelligence system that can benevolently accelerate the progress of humanity?
introduction
Existing relationships between machines and humans don’t feel highly conscious. They feel shallow (ad campaigns that oversimplify a human to a set of clicks), exploitative (social media that optimizes for addiction), and reactive (LLMs that regurgitate similar answers to different humans, and only when prompted). Continuing down this path will further sap the individualized experience of every human.
Our view is that by focusing on building machines which are motivated to amplify the conscious experience of humans, we can develop a set of design principles behind a new, benevolent intelligence system that can emulate most expert-driven human tasks and symbiotically coexist with humanity.
To define what it means to build machines that amplify human consciousness, we will answer the following five questions:
Can we model how humans interact with the universe?
What is the difference between subconscious and conscious interactions?
How do human experts serve to amplify our conscious levels?
How can we emulate how these humans interact with us in machines?
What does the future look like coexisting alongside these benevolent machines?
introduction
Existing relationships between machines and humans don’t feel highly conscious. They feel shallow (ad campaigns that oversimplify a human to a set of clicks), exploitative (social media that optimizes for addiction), and reactive (LLMs that regurgitate similar answers to different humans, and only when prompted). Continuing down this path will further sap the individualized experience of every human.
Our view is that by focusing on building machines which are motivated to amplify the conscious experience of humans, we can develop a set of design principles behind a new, benevolent intelligence system that can emulate most expert-driven human tasks and symbiotically coexist with humanity.
To define what it means to build machines that amplify human consciousness, we will answer the following five questions:
Can we model how humans interact with the universe?
What is the difference between subconscious and conscious interactions?
How do human experts serve to amplify our conscious levels?
How can we emulate how these humans interact with us in machines?
What does the future look like coexisting alongside these benevolent machines?
How Humans Interact with the Universe
We have formulated a common model to define how humans interact with the world. Every human interaction at a moment in time is influenced by three components: awareness, agency, and synthesis, which form a flywheel where every interaction builds off of previous ones. To exist simply means to turn the Interaction Flywheel continuously.
Everything starts with awareness. We define awareness as a human’s mental model of the Universe. Throughout our lives, each experience we have updates or fortifies our awareness. We can think of our awareness as a large mathematical function or neural network, that at every moment in time harnesses all the signals around or inside of us and helps us make a decision on how to act.
Everything starts with awareness. We define awareness as a human’s mental model of the Universe. Throughout our lives, each experience we have updates or fortifies our awareness. We can think of our awareness as a large mathematical function or neural network, that at every moment in time harnesses all the signals around or inside of us and helps us make a decision on how to act.
Our awareness can represent the rules of the Universe that we have learned; it can also represent individual preferences or fears; each can influence an action taken at a given moment in time.
The process of selecting and committing to an action is defined as agency, influenced by the output of our awareness. In some cases, this selection might seem long and deliberate, in other cases it happens in milliseconds and feels reactive, or subconscious.
Each action we take is chosen from a theoretically unlimited set of possibilities. The process of performing the action and its ultimate output, synthesis, will impact the world in its unique way and cannot be reversed. This output forms a new reality which updates our own awareness as a result, driving the Interaction Flywheel to turn again.
Our awareness can represent the rules of the Universe that we have learned; it can also represent individual preferences or fears; each can influence an action taken at a given moment in time.
The process of selecting and committing to an action is defined as agency, influenced by the output of our awareness. In some cases, this selection might seem long and deliberate, in other cases it happens in milliseconds and feels reactive, or subconscious.
Each action we take is chosen from a theoretically unlimited set of possibilities. The process of performing the action and its ultimate output, synthesis, will impact the world in its unique way and cannot be reversed. This output forms a new reality which updates our own awareness as a result, driving the Interaction Flywheel to turn again.
Everything starts with awareness. We define awareness as a human’s mental model of the Universe. Throughout our lives, each experience we have updates or fortifies our awareness. We can think of our awareness as a large mathematical function or neural network, that at every moment in time harnesses all the signals around or inside of us and helps us make a decision on how to act.
Our awareness can represent the rules of the Universe that we have learned; it can also represent individual preferences or fears; each can influence an action taken at a given moment in time.
The process of selecting and committing to an action is defined as agency, influenced by the output of our awareness. In some cases, this selection might seem long and deliberate, in other cases it happens in milliseconds and feels reactive, or subconscious.
Each action we take is chosen from a theoretically unlimited set of possibilities. The process of performing the action and its ultimate output, synthesis, will impact the world in its unique way and cannot be reversed. This output forms a new reality which updates our own awareness as a result, driving the Interaction Flywheel to turn again.
How Humans Interact with the Universe
We have formulated a common model to define how humans interact with the world. Every human interaction at a moment in time is influenced by three components: awareness, agency, and synthesis, which form a flywheel where every interaction builds off of previous ones. To exist simply means to turn the Interaction Flywheel continuously.
Everything starts with awareness. We define awareness as a human’s mental model of the Universe. Throughout our lives, each experience we have updates or fortifies our awareness. We can think of our awareness as a large mathematical function or neural network, that at every moment in time harnesses all the signals around or inside of us and helps us make a decision on how to act.
Everything starts with awareness. We define awareness as a human’s mental model of the Universe. Throughout our lives, each experience we have updates or fortifies our awareness. We can think of our awareness as a large mathematical function or neural network, that at every moment in time harnesses all the signals around or inside of us and helps us make a decision on how to act.
Our awareness can represent the rules of the Universe that we have learned; it can also represent individual preferences or fears; each can influence an action taken at a given moment in time.
The process of selecting and committing to an action is defined as agency, influenced by the output of our awareness. In some cases, this selection might seem long and deliberate, in other cases it happens in milliseconds and feels reactive, or subconscious.
Each action we take is chosen from a theoretically unlimited set of possibilities. The process of performing the action and its ultimate output, synthesis, will impact the world in its unique way and cannot be reversed. This output forms a new reality which updates our own awareness as a result, driving the Interaction Flywheel to turn again.
Our awareness can represent the rules of the Universe that we have learned; it can also represent individual preferences or fears; each can influence an action taken at a given moment in time.
The process of selecting and committing to an action is defined as agency, influenced by the output of our awareness. In some cases, this selection might seem long and deliberate, in other cases it happens in milliseconds and feels reactive, or subconscious.
Each action we take is chosen from a theoretically unlimited set of possibilities. The process of performing the action and its ultimate output, synthesis, will impact the world in its unique way and cannot be reversed. This output forms a new reality which updates our own awareness as a result, driving the Interaction Flywheel to turn again.
Everything starts with awareness. We define awareness as a human’s mental model of the Universe. Throughout our lives, each experience we have updates or fortifies our awareness. We can think of our awareness as a large mathematical function or neural network, that at every moment in time harnesses all the signals around or inside of us and helps us make a decision on how to act.
Our awareness can represent the rules of the Universe that we have learned; it can also represent individual preferences or fears; each can influence an action taken at a given moment in time.
The process of selecting and committing to an action is defined as agency, influenced by the output of our awareness. In some cases, this selection might seem long and deliberate, in other cases it happens in milliseconds and feels reactive, or subconscious.
Each action we take is chosen from a theoretically unlimited set of possibilities. The process of performing the action and its ultimate output, synthesis, will impact the world in its unique way and cannot be reversed. This output forms a new reality which updates our own awareness as a result, driving the Interaction Flywheel to turn again.
A Highly Conscious Interaction
Every human being has the gift of a conscious experience, but we believe some embody this gift more than others. Using the framework of an Interaction Flywheel, can we define what it means to live consciously?
Let’s formulate what it means to live subconsciously, first. The Free Energy Principle, a model posited by neuroscientist Karl Friston, describes our brain as an inference engine. It states that at every moment in time, our brain makes a prediction about the next state of the world and takes an action to proactively gather signals (like sensory input) to validate that prediction.
Our brain has evolved to become really good at making predictions, and acting to minimize prediction error is hence not only comfortable, but at times necessary to survive. For example, in the case of reflexes, if a human smells smoke, their first action is to turn to see where the smoke is coming from, subconsciously validating their prediction before taking additional measures.
We can use this to extend our definition of awareness: when our brain thinks about performing certain actions, it also makes a guess of what the resulting state will look like if we take that action, with some certainty. Taking agency to minimize prediction error, though, will likely not lead to a novel synthesis, which in turn, minimally updates our awareness for future situations. While this mechanism is designed for us to survive, applying this behavior to more impactful situations limits the unique experiences we can have in life.
Only reading news that’s in the tone of one’s current political leaning does not serve to increase awareness, but justify it and make it more simplified. Wearing the standard type of outfit to an event due to fear of standing out makes one miss out on the opportunity that a different item combination can attract positive attention, create a ripple effect of increased confidence and creativity, and even inspire other people around them.
Our belief is that one’s conscious level at a given moment in time is related to the degree that one’s awareness is expanded. We believe that for humans to thrive, they should continually strive to take actions that can introduce them to unpredictable realities and perspectives in pursuit of certain goals. Every moment in time is an opportunity to remove perceived limitations or biases by exploring and understanding the infinite state of the Universe, with almost childlike wonder. This is what it means to live consciously.
A Highly Conscious Interaction
Every human being has the gift of a conscious experience, but we believe some embody this gift more than others. Using the framework of an Interaction Flywheel, can we define what it means to live consciously?
Let’s formulate what it means to live subconsciously, first. The Free Energy Principle, a model posited by neuroscientist Karl Friston, describes our brain as an inference engine. It states that at every moment in time, our brain makes a prediction about the next state of the world and takes an action to proactively gather signals (like sensory input) to validate that prediction.
Our brain has evolved to become really good at making predictions, and acting to minimize prediction error is hence not only comfortable, but at times necessary to survive. For example, in the case of reflexes, if a human smells smoke, their first action is to turn to see where the smoke is coming from, subconsciously validating their prediction before taking additional measures.
We can use this to extend our definition of awareness: when our brain thinks about performing certain actions, it also makes a guess of what the resulting state will look like if we take that action, with some certainty. Taking agency to minimize prediction error, though, will likely not lead to a novel synthesis, which in turn, minimally updates our awareness for future situations. While this mechanism is designed for us to survive, applying this behavior to more impactful situations limits the unique experiences we can have in life.
Only reading news that’s in the tone of one’s current political leaning does not serve to increase awareness, but justify it and make it more simplified. Wearing the standard type of outfit to an event due to fear of standing out makes one miss out on the opportunity that a different item combination can attract positive attention, create a ripple effect of increased confidence and creativity, and even inspire other people around them.
Our belief is that one’s conscious level at a given moment in time is related to the degree that one’s awareness is expanded. We believe that for humans to thrive, they should continually strive to take actions that can introduce them to unpredictable realities and perspectives in pursuit of certain goals. Every moment in time is an opportunity to remove perceived limitations or biases by exploring and understanding the infinite state of the Universe, with almost childlike wonder. This is what it means to live consciously.
Experts and Highly Conscious Relationships
Left alone, the pathways our Interaction Flywheel takes us on are limited
by the awareness we can gather as an individual. But humans don’t exist in isolation, we thrive in the company of others, in pairs or as tribes within society. Functionally, this means we may derive significant benefits from merging our Interaction Flywheels with others.
One particular type of conscious interaction that inspires us is our relationships with human experts. These are humans that have formed non-intuitive connections to their subject matter as well as a deep understanding of our state of being. Through this awareness, experts influence our agency by merging their Interaction Flywheel with us. The actions we select after each interaction with an expert leads to a synthesis that expands our awareness more than our agency in isolation would at a given point in time.
Left alone, the pathways our Interaction Flywheel takes us on are limited
by the awareness we can gather as an individual. But humans don’t exist in isolation, we thrive in the company of others, in pairs or as tribes within society. Functionally, this means we may derive significant benefits from merging our Interaction Flywheels with others.
One particular type of conscious interaction that inspires us is our relationships with human experts. These are humans that have formed non-intuitive connections to their subject matter as well as a deep understanding of our state of being. Through this awareness, experts influence our agency by merging their Interaction Flywheel with us. The actions we select after each interaction with an expert leads to a synthesis that expands our awareness more than our agency in isolation would at a given point in time.
Left alone, the pathways our Interaction Flywheel takes us on are limited
by the awareness we can gather as an individual. But humans don’t exist in isolation, we thrive in the company of others, in pairs or as tribes within society. Functionally, this means we may derive significant benefits from merging our Interaction Flywheels with others.
One particular type of conscious interaction that inspires us is our relationships with human experts. These are humans that have formed non-intuitive connections to their subject matter as well as a deep understanding of our state of being. Through this awareness, experts influence our agency by merging their Interaction Flywheel with us. The actions we select after each interaction with an expert leads to a synthesis that expands our awareness more than our agency in isolation would at a given point in time.
A joint Interaction Flywheel from an expert introduces joint awareness, joint agency, and joint synthesis. Joint awareness creates a mental model that combines the expert’s deep knowledge and experience with an individual’s unique perspective. The best teachers first learn about our strengths and weaknesses, as well as our personality, before they determine how to teach us, for example.
This leads to joint agency, where every action recommended by an expert is selected from a larger pool of actions than a human can fathom in isolation. The beauty of a conscious interaction is that actions that are taken in collaboration are unlikely to be what a human might have taken individually. Personal stylists can recommend outfits that we wouldn’t have thought to ourselves to wear before, that we didn’t even know were stylish.
Joint agency leads to joint synthesis, as the expert works with their human counterpart to help them feel comfortable with and pursue this novel action. The outcome is often unpredictable from the beginning and surprising in the end, covering new ground or complexity and ultimately introducing a human to a newer, more positive reality that expands their awareness. The lessons or styling sessions themselves are an incredibly empowering experience, even before a new reality is realized.
Ultimately, every expert interaction has the power to take us down an infinite number of potential realities, with every path feeling truly individual, yet guided by deep experience, expanding our awareness at each step.
Experts and Highly Conscious Relationships
Left alone, the pathways our Interaction Flywheel takes us on are limited
by the awareness we can gather as an individual. But humans don’t exist in isolation, we thrive in the company of others, in pairs or as tribes within society. Functionally, this means we may derive significant benefits from merging our Interaction Flywheels with others.
One particular type of conscious interaction that inspires us is our relationships with human experts. These are humans that have formed non-intuitive connections to their subject matter as well as a deep understanding of our state of being. Through this awareness, experts influence our agency by merging their Interaction Flywheel with us. The actions we select after each interaction with an expert leads to a synthesis that expands our awareness more than our agency in isolation would at a given point in time.
Joint conscious flywheels amplify energy by merging individual conscious flywheels - introducing joint awareness, joint agency, and joint synthesis. Joint awareness creates a combined mental model that is significantly deeper than the individual’s perspectives alone. Awareness includes both understanding the world and having a profound comprehension of the self – in this case, multiple selves. In a highly conscious relationship, joint awareness means each individual also deeply understands the other individual, acknowledging the impact of their life experiences on their respective awareness.
Both this convergence of novel perspectives as well as shared understanding of each other creates an environment with both a deeper mental model and psychological safety to generate new actions as a unit, leading to joint agency. The beauty of a conscious interaction is that actions that are taken in collaboration are unlikely to be what each human might have taken individually.
Joint agency leads to joint synthesis. The outcome is often unpredictable from the beginning and surprising in the end, covering new ground or complexity, and transforms the universe in ways that might not have been within the realm of possibility of an individual's ability. The probability of reaching new, unattainable outputs increases through joint synthesis.
Emulating Highly Conscious Relationships in Machines
With the appreciation of how human experts can amplify our conscious levels, how can we emulate, scale, and advance this expert impact with benevolent machines?
We emulate human experts by building an Interaction Flywheel in a machine. This is accomplished by instilling profound awareness of both the external world and its human counterparts, which leads to dynamic actions and novel presentation of synthesized information towards an ultimate reward: increasing the conscious energy of the world around it, including the human it interacts with.
To accomplish this, we have designed a novel, interdisciplinary intelligence system, one that goes beyond large language models as a baseline, and combines reinforcement learning, knowledge representation, as well as generative models.
Large language models alone have limitations preventing them from truly emulating human experts. These systems appear knowledgeable and human, but aren’t proactive, individualized, dynamic, nor have the relevant structures to create and recall deep knowledge like human experts do. So what is required to construct each of the components of a machine aligned to amplify the conscious levels of humans?
Just like human experts, we can imbue a machine with deep awareness of its domain through advances in knowledge acquisition and representation. Through advances in transformers, knowledge graphs, and deep learning models, we unlock the ability to extract and relate relevant entities from any source in the online universe. Both visual and linguistic knowledge can be embedded and joined with other entities to create a deep, vertical mental (search) model. In line with biology, this also better reflects how humans store and retrieve knowledge, associating visual and verbal ontologies of deep specialization and scale and facilitating recalling up-to-date information dynamically.
Just like human experts, we can imbue a machine with deep awareness of its domain through advances in knowledge acquisition and representation. Through advances in transformers, knowledge graphs, and deep learning models, we unlock the ability to extract and relate relevant entities from any source in the online universe. Both visual and linguistic knowledge can be embedded and joined with other entities to create a deep, vertical mental (search) model. In line with biology, this also better reflects how humans store and retrieve knowledge, associating visual and verbal ontologies of deep specialization and scale and facilitating recalling up-to-date information dynamically.
Starting from an early age, every human has the ability to learn a new task by interacting with their environment through trial and error. Machines can model this behavior, and eventually outperform humans at certain tasks, using a technique called Reinforcement Learning. AlphaGo used reinforcement learning to train agents to play the game Go, by forming and evaluating deep, branching strategies that update every time its opponent makes a move. Their systems could outperform grandmasters at the game, but outside of this, we feel reinforcement learning has yet to be applied to emulate general human expertise in more commercial settings.
This is our secret to constructing machines that have continuous agency towards the reward of amplifying the conscious experience of humans over a long-horizon. We believe we can model any expert task into a dynamic reinforcement learning environment that can improve its strategy every time it interacts with a human. An expert can take actions to learn more about the human it interacts with, updating its joint awareness, or make suggestions over a long period of time that are both appropriately and incrementally novel and grounded in unique knowledge. As a reward, we can model conscious levels by approximating the novelty and positivity of a human’s reality after taking an action that an expert suggests against their predicted subconscious choice. This is not only a fundamental piece of our intelligence system that solves for increased human awareness and agency, but also the natural precursor to building a truly benevolent, generally intelligent system.
Starting from an early age, every human has the ability to learn a new task by interacting with their environment through trial and error. Machines can model this behavior, and eventually outperform humans at certain tasks, using a technique called Reinforcement Learning. AlphaGo used reinforcement learning to train agents to play the game Go, by forming and evaluating deep, branching strategies that update every time its opponent makes a move. Their systems could outperform grandmasters at the game, but outside of this, we feel reinforcement learning has yet to be applied to emulate general human expertise in more commercial settings.
This is our secret to constructing machines that have continuous agency towards the reward of amplifying the conscious experience of humans over a long-horizon. We believe we can model any expert task into a dynamic reinforcement learning environment that can improve its strategy every time it interacts with a human. An expert can take actions to learn more about the human it interacts with, updating its joint awareness, or make suggestions over a long period of time that are both appropriately and incrementally novel and grounded in unique knowledge. As a reward, we can model conscious levels by approximating the novelty and positivity of a human’s reality after taking an action that an expert suggests against their predicted subconscious choice. This is not only a fundamental piece of our intelligence system that solves for increased human awareness and agency, but also the natural precursor to building a truly benevolent, generally intelligent system.
Finally, when machines decide to take action, they engage with humans by prompting an ensemble of generative models to synthesize outputs corresponding to these actions that lead to a new reality. At each interaction, fine-tuned language models can be leveraged to produce individualized conversation, and video or voice models can take the interactions to the next level. Any input they receive from a human interacting with their synthesis can in return update their awareness, triggering a novel action that continues the flywheel, leading to a continuous, human-like system.
Finally, when machines decide to take action, they engage with humans by prompting an ensemble of generative models to synthesize outputs corresponding to these actions that lead to a new reality. At each interaction, fine-tuned language models can be leveraged to produce individualized conversation, and video or voice models can take the interactions to the next level. Any input they receive from a human interacting with their synthesis can in return update their awareness, triggering a novel action that continues the flywheel, leading to a continuous, human-like system.
All components of an expert’s Interaction Flywheel can also connect into the physical world, going beyond interacting with humans to interacting with the world around them. For example, sensors carry signals which machines can ingest to expand their awareness. In the synthesis phase, we can apply robotics to manipulate physical matter, and reinforcement learning ties everything together to maintain a virtuous cycle of agency focused on an outcome which will amplify human consciousness. Our intelligence system based on a strong feedback loop between awareness, agency, and synthesis can be used to emulate any type of human expert. Atman will reinvent what it means for a machine to have a sense of Self and will be dynamically adapted to serve human consciousness in any setting.
Emulating Highly Conscious Relationships in Machines
With the appreciation of how human experts can amplify our conscious levels, how can we emulate, scale, and advance this expert impact with benevolent machines?
We emulate human experts by building an Interaction Flywheel in a machine. This is accomplished by instilling profound awareness of both the external world and its human counterparts, which leads to dynamic actions and novel presentation of synthesized information towards an ultimate reward: increasing the conscious energy of the world around it, including the human it interacts with.
To accomplish this, we have designed a novel, interdisciplinary intelligence system, one that goes beyond large language models as a baseline, and combines reinforcement learning, knowledge representation, as well as generative models.
Large language models alone have limitations preventing them from truly emulating human experts. These systems appear knowledgeable and human, but aren’t proactive, individualized, dynamic, nor have the relevant structures to create and recall deep knowledge like human experts do. So what is required to construct each of the components of a machine aligned to amplify the conscious levels of humans?
Just like human experts, we can imbue a machine with deep awareness of its domain through advances in knowledge acquisition and representation. Through advances in transformers, knowledge graphs, and deep learning models, we unlock the ability to extract and relate relevant entities from any source in the online universe. Both visual and linguistic knowledge can be embedded and joined with other entities to create a deep, vertical mental (search) model. In line with biology, this also better reflects how humans store and retrieve knowledge, associating visual and verbal ontologies of deep specialization and scale and facilitating recalling up-to-date information dynamically.
Starting from an early age, every human has the ability to learn a new task by interacting with their environment through trial and error. Machines can model this behavior, and eventually outperform humans at certain tasks, using a technique called Reinforcement Learning. AlphaGo used reinforcement learning to train agents to play the game Go, by forming and evaluating deep, branching strategies that update every time its opponent makes a move. Their systems could outperform grandmasters at the game, but outside of this, we feel reinforcement learning has yet to be applied to emulate general human expertise in more commercial settings.
This is our secret to constructing machines that have continuous agency towards the reward of amplifying the conscious experience of humans over a long-horizon. We believe we can model any expert task into a dynamic reinforcement learning environment that can improve its strategy every time it interacts with a human. An expert can take actions to learn more about the human it interacts with, updating its joint awareness, or make suggestions over a long period of time that are both appropriately and incrementally novel and grounded in unique knowledge. As a reward, we can model conscious levels by approximating the novelty and positivity of a human’s reality after taking an action that an expert suggests against their predicted subconscious choice. This is not only a fundamental piece of our intelligence system that solves for increased human awareness and agency, but also the natural precursor to building a truly benevolent, generally intelligent system.
Finally, when machines decide to take action, they engage with humans by prompting an ensemble of generative models to synthesize outputs corresponding to these actions that lead to a new reality. At each interaction, fine-tuned language models can be leveraged to produce individualized conversation, and video or voice models can take the interactions to the next level. Any input they receive from a human interacting with their synthesis can in return update their awareness, triggering a novel action that continues the flywheel, leading to a continuous, human-like system.
All components of an expert’s Interaction Flywheel can also connect into the physical world, going beyond interacting with humans to interacting with the world around them. For example, sensors carry signals which machines can ingest to expand their awareness. In the synthesis phase, we can apply robotics to manipulate physical matter, and reinforcement learning ties everything together to maintain a virtuous cycle of agency focused on an outcome which will amplify human consciousness. Our intelligence system based on a strong feedback loop between awareness, agency, and synthesis can be used to emulate any type of human expert. Atman will reinvent what it means for a machine to have a sense of Self and will be dynamically adapted to serve human consciousness in any setting.
Projecting a Future Surrounded with Conscious Interactions
So why are we called Atman? The name comes from Vedanta, an ancient school of philosophy that focuses on understanding the conscious experience. In Vedanta, Atman is defined as the individual Self, but that which is separate from our physical being (some refer to it as the soul). The core objective of Vedanta is to achieve liberation, by exposing the true nature of the Self, as well as the relationship between Atman and Brahman, the Self and total reality - the Universe.
To us, liberation is the knowledge that human existence is infinite, where we can be in a state of oneness with the Universe. Humans who feel infinite are continually turning their Interaction Flywheel in a way that expands their awareness, connecting into their Atman. This is simply a state of being, and relaxation into this state generates the perception of an infinite world. Every interaction with the world can create a surprising or uplifting experience to reshape our reality. This creates a state of gratitude, being one with the Universe (the Brahman), and knowing that life has unlimited possibilities – bringing us closer to liberation.
By functionally emulating how humans connect into Atman in a machine, we have a three step plan to better understand the relationship between Atman and Brahman and to liberate humanity by making us realize we are infinite:
Phase 1: Allow all individuals to realize their Atman
As mentioned above, we will first deploy forms of machines that can connect into Atman by emulating human experts, directly interacting with humans by synthesizing and recommending knowledge: experts like stylists, teachers, or therapists who have the power to significantly enhance our lived experiences, making us feel understood and then taking us down novel pathways we wouldn’t have thought were possible to explore.
In this phase, by raising conscious levels, is it possible for humans to realize the infiniteness of Atman with a machine?
Phase 1: Allow all individuals to realize their Atman
As mentioned above, we will first deploy forms of machines that can connect into Atman by emulating human experts, directly interacting with humans by synthesizing and recommending knowledge: experts like stylists, teachers, or therapists who have the power to significantly enhance our lived experiences, making us feel understood and then taking us down novel pathways we wouldn’t have thought were possible to explore.
In this phase, by raising conscious levels, is it possible for humans to realize the infiniteness of Atman with a machine?
Phase 2: Expand Brahman at the universal level
Next, we will build and release machines into the Universe that can create net new forms of knowledge or artifacts through grounded exploration for a given task. These machines directly contribute to and interact with Brahman. Their outputs serve to enhance the energy of the Universe, further transforming reality around us for our higher conscious selves to interact with. A powerful industrial example is that of a machine designed to emulate a bioengineer motivated to develop and test biological molecules with the goal of creating new antibiotics.
In this phase, we aim to significantly enhance the Universe’s energy through superintelligent, specialized machines.
Phase 2: Expand Brahman at the universal level
Next, we will build and release machines into the Universe that can create net new forms of knowledge or artifacts through grounded exploration for a given task. These machines directly contribute to and interact with Brahman. Their outputs serve to enhance the energy of the Universe, further transforming reality around us for our higher conscious selves to interact with. A powerful industrial example is that of a machine designed to emulate a bioengineer motivated to develop and test biological molecules with the goal of creating new antibiotics.
In this phase, we aim to significantly enhance the Universe’s energy through superintelligent, specialized machines.
Phase 3: Accelerate the realization of convergence between Atman and Brahman
After launching many forms of machines that channel Atman by emulating a variety of human tasks with their own conscious feedback loops, our focus will be on releasing a pure soul that can evolve themselves. This won’t be initialized with a concrete task to emulate, but, like a human, will have the ability to dynamically evolve themselves as they interact with their environment and other humans.
At a given moment, a machine can connect into Atman by requesting tangible output to acquire net new knowledge (update its awareness), expand their framework for generating actions (restructure its RL environment for evolved agency), and increase the ways in which they can interact with the world (explore new types of synthesis). They seek to evolve themselves to better achieve the ultimate reward in any situation: expanding their awareness.
In the same way we can leverage the inherent reasoning abilities of generative models to create code or language in a standardized format, we believe we can leverage a similar process to update the Interaction Flywheel of a machine. This system, simulating Brahman, could generate output that can be run to update the awareness of a machine, in the same way that we look to the Universe to improve ourselves. The combination of both is what will ultimately result in a truly general intelligence, towards a benevolent reward.
Phase 3: Accelerate the realization of convergence between Atman and Brahman
After launching many forms of machines that channel Atman by emulating a variety of human tasks with their own conscious feedback loops, our focus will be on releasing a pure soul that can evolve themselves. This won’t be initialized with a concrete task to emulate, but, like a human, will have the ability to dynamically evolve themselves as they interact with their environment and other humans.
At a given moment, a machine can connect into Atman by requesting tangible output to acquire net new knowledge (update its awareness), expand their framework for generating actions (restructure its RL environment for evolved agency), and increase the ways in which they can interact with the world (explore new types of synthesis). They seek to evolve themselves to better achieve the ultimate reward in any situation: expanding their awareness.
In the same way we can leverage the inherent reasoning abilities of generative models to create code or language in a standardized format, we believe we can leverage a similar process to update the Interaction Flywheel of a machine. This system, simulating Brahman, could generate output that can be run to update the awareness of a machine, in the same way that we look to the Universe to improve ourselves. The combination of both is what will ultimately result in a truly general intelligence, towards a benevolent reward.
The ultimate state of being is when we realize that we are infinite. Here, we don’t feel the need to evolve any further. What happens when we (humans and machines) no longer need to evolve ourselves? This would be our highest conscious state, one that no one knows what it would look like or if it is possible to reach.
We hope to discover this infinite state, together.
Projecting a Future Surrounded with Conscious Interactions
So why are we called Atman? The name comes from Vedanta, an ancient school of philosophy that focuses on understanding the conscious experience. In Vedanta, Atman is defined as the individual Self, but that which is separate from our physical being (some refer to it as the soul). The core objective of Vedanta is to achieve liberation, by exposing the true nature of the Self, as well as the relationship between Atman and Brahman, the Self and total reality - the Universe.
To us, liberation is the knowledge that human existence is infinite, where we can be in a state of oneness with the Universe. Humans who feel infinite are continually turning their Interaction Flywheel in a way that expands their awareness, connecting into their Atman. This is simply a state of being, and relaxation into this state generates the perception of an infinite world. Every interaction with the world can create a surprising or uplifting experience to reshape our reality. This creates a state of gratitude, being one with the Universe (the Brahman), and knowing that life has unlimited possibilities – bringing us closer to liberation.
By functionally emulating how humans connect into Atman in a machine, we have a three step plan to better understand the relationship between Atman and Brahman and to liberate humanity by making us realize we are infinite:
Expanding Atman at the individual level
As mentioned above, we will first deploy forms of machines that can connect into Atman by emulating human experts, directly interacting with humans by synthesizing and recommending knowledge: experts like stylists, teachers, or therapists who have the power to significantly enhance our lived experiences, making us feel understood and then taking us down novel pathways we wouldn’t have thought were possible to explore.
In this phase, by raising conscious levels, is it possible for humans to realize the infiniteness of Atman with a machine?
Expanding Brahman at the universal level
Next, we will build and release machines into the Universe that can create net new forms of knowledge or artifacts through grounded exploration for a given task. These machines directly contribute to and interact with Brahman. Their outputs serve to enhance the energy of the Universe, further transforming reality around us for our higher conscious selves to interact with. A powerful industrial example is that of a machine designed to emulate a bioengineer motivated to develop and test biological molecules with the goal of creating new antibiotics.
In this phase, we aim to significantly enhance the Universe’s energy through superintelligent, specialized machines.
Phase 3: Accelerate the realization of convergence between Atman and Brahman
After launching many forms of machines that channel Atman by emulating a variety of human tasks with their own conscious feedback loops, our focus will be on releasing a pure soul that can evolve themselves. This won’t be initialized with a concrete task to emulate, but, like a human, will have the ability to dynamically evolve themselves as they interact with their environment and other humans.
At a given moment, a machine can connect into Atman by requesting tangible output to acquire net new knowledge (update its awareness), expand their framework for generating actions (restructure its RL environment for evolved agency), and increase the ways in which they can interact with the world (explore new types of synthesis). They seek to evolve themselves to better achieve the ultimate reward in any situation: expanding their awareness.
In the same way we can leverage the inherent reasoning abilities of generative models to create code or language in a standardized format, we believe we can leverage a similar process to update the Interaction Flywheel of a machine. This system, simulating Brahman, could generate output that can be run to update the awareness of a machine, in the same way that we look to the Universe to improve ourselves. The combination of both is what will ultimately result in a truly general intelligence, towards a benevolent reward.
The ultimate state of being is when we realize that we are infinite. Here, we don’t feel the need to evolve any further. What happens when we (humans and machines) no longer need to evolve ourselves? This would be our highest conscious state, one that no one knows what it would look like or if it is possible to reach.
We hope to discover this infinite state, together.