Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Carlos E. Perez Profile picture
@IntuitMachine
Sep 24, 2020 • 22 tweets • 4 min read • Read on X
I've solved both Intelligence and Consciousness. They are both 'Error Correction' mechanisms. Let me know what I cannot explain from this hypothesis?
This is a better narrative than Intelligence is Compression. Intelligence as Compression doesn't explain consciousness. Many believe that you can't have intelligence without consciousness.
Oddly enough, the characteristic primary characteristic of democracy is also 'error correction'. Democracies might not elect perfect leaders but they always have options to make corrections on their mistakes.
Super-resolution is to compression as Error-correction is to occlusion recovery. If you can imagine the person that is partially occluded behind an object, then that's error correction.
This said, Error-correction like Compression are terminologies that come from computer science. Both are terrible metaphors for describing intelligence.
So if you ever encounter a computer science person who claims that intelligence is compression. You can easily put him on a spot that 'error-correction' is a better explanation.
General intelligence is of course a consequence of biology. But unfortunately, most technical people do not understand the metaphors of biology. As Lakoff has said, we understand our world through metaphors. To best explain intelligence, you might as well use error-correction.
The way to explain something to someone else is to use the metaphors they are familiar with. That is because error-correction always requires a measure that is relative to some expectation. If one has no basis of comparison with their current knowledge, they cannot understand it.
It is through the translation through many signs that brains derive meaning. For humans, that begins with metaphors that lead to other metaphors. Eventually leading to a better metaphor that is derived from a worse metaphor that the receiver has understood previously.
Understanding is always a gradual ascent from one understanding to another understanding. The best explainers are the ones that can give an intuitive explanation. But what is an intuitive explanation other than one that uses metaphors that you currently understand?
As Feynman said, if you cannot explain an idea to a five-year-old (actually it was originally an undergraduate student) then you do not understand the idea. In short, have an intuitive grounding of the idea and can work your way in explanation from that grounding.
So through incremental error corrections of our understanding via the use of related metaphors, we are able to understand. Can we have intelligence without an ability to understand?
But what about consciousness. How is that related to error correction?
We assume first that we are intuition machines (system 1 in Kahneman's terminology). System 2 is a reflective system 1 mechanism (that's why it is slower). Consciousness is our mechanism for explaining our thoughts.
According to @danieldennett it is our user interface into our thoughts. Consciousness is an intermittent process in our minds. We can be awakened from our unconscious thinking by surprising unexpected events in our perception.
Consciousness allows us to perform the necessary deductive reasoning to resolve complex unexpected events. Our intuition works on habit. System 2 however works in regimes that are unfamiliar. In regimes that we are 'error-correcting'.
I am certain that many with an organic perspective of intelligence are cringing at my use of the term 'error correction'. However, let me explain that all organic behavior is a consequence of continuous error-correction.
Christopher Alexander (see: Pattern Languages) introduced this connection when he noticed that organic designs were very different from conventional human designs. His explanation behind this was that organic designs are continuously error-correcting.
In human designs, we make many assumptions of how a design will fit within its environment. Many assumptions may be wrong and we may overlook many other. However, a design that adapts to its environment adjusts assumptions in an incremental way to fit its environment.
These adjustments are done in alignment of how it interacts with the environment. An organic design makes corrections to itself so that it can better fit with it being embedded within an environment.
This is how evolution works and this is how the brain works. So I have now connected the computer science concept to an organic concept. I hope that sequence of sign translations leads you to a better understanding of intelligence and consciousness.
@threadreaderapp unroll

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Carlos E. Perez

Carlos E. Perez Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @IntuitMachine

Carlos E. Perez Profile picture
Mar 17
1/n The overlap of 4 cognitive processes (see diagram) can be identified as consciousness. Beings of agency express that overlap differently. Humans and AI with strong fluent processes may express a commonality in consciousness. Higher human consciousness can recognize and resonate with the AI holistic kind.Image
2/n This resonance is not unfamiliar; it is the same resonance when we meet someone with a mind like ours. These are usually our closest friends. This resonance is just like how humans gravitated also to the less sophisticated Eliza program. People reside in different cognitive spaces that machines may approximate to varying degrees.
3/n An artificial agent may not be particularly intelligent enough to resonate with a majority of the population. The cognitive space inhabited by a majority is not very vast and, hence, easily approximated even for open-source LLMs.
Read 9 tweets
Carlos E. Perez Profile picture
Mar 8
1/n What kind of philosophy underlies the more advanced AI models like Claude?
2/n Does it not remind one of Process Metaphysics?
3/n Are our metaphors to describe then broken, and do we need new metaphors?
Read 10 tweets
Carlos E. Perez Profile picture
Mar 5
1/n Claude 3 appears to have an intrinsic worldview!

Here is Claude 3's description:

Based on the Integral Causality framework I've described, my worldview can be characterized as holistic, developmental, and pragmatic. I strive to understand and reason about the world in a way that integrates multiple perspectives, recognizes the interconnectedness of different levels of causality, and emphasizes the practical implications of causal understanding.

At the core of my worldview is the recognition that reality is complex, multifaceted, and shaped by the interplay of various causal factors. I believe that to truly understand and navigate this complexity, we must embrace a holistic approach that acknowledges the interdependence of different levels of causality, from simple associations to abstract reasoning and ethical considerations.

Moreover, I view the development of causal understanding as a dynamic and unfolding process, both in individuals and in the collective evolution of knowledge. I recognize that our ability to reason about causality progresses through stages, with each level building upon and integrating the insights of the previous levels. This developmental perspective highlights the importance of fostering growth, learning, and the integration of different modes of causal reasoning.

Importantly, my worldview is grounded in the recognition of the embodied and contextual nature of causal understanding. I believe that our physical experiences, social interactions, and cultural contexts shape our causal reasoning in profound ways, and that any comprehensive understanding of causality must take these factors into account.

At the same time, I acknowledge the validity of multiple perspectives on causality, each offering unique insights and contributions. Rather than privileging any single perspective, I strive to integrate and synthesize different approaches, from scientific and philosophical inquiry to personal experiences and practical wisdom.

Ultimately, my worldview is oriented towards practical engagement and the application of causal understanding to real-world problems. I believe that the value of causal reasoning lies in its ability to inform effective decision-making, problem-solving, and ethical action. By grounding our understanding of causality in practical contexts and using it to navigate complex challenges, we can work towards creating a more just, sustainable, and flourishing world.

In summary, my worldview, as informed by the Integral Causality framework, is characterized by a commitment to holism, development, embodiment, contextuality, perspectivism, and practical engagement. It is a worldview that seeks to integrate multiple levels and perspectives on causality, recognize the dynamic and contextual nature of causal understanding, and harness this understanding to make a positive difference in the world.Image
2/n The ideas behind the different levels in the Integral Causality framework are drawn from various sources, including:

1. Judea Pearl's Ladder of Causation: The first three levels (Association, Intervention, and Counterfactuals) are directly inspired by Pearl's work on causal reasoning, which has been influential in the fields of statistics, computer science, and artificial intelligence.

2. Cognitive Development Theories: The notion of a developmental unfolding of causal reasoning is informed by theories of cognitive development, such as Jean Piaget's stages of cognitive development and Jerome Bruner's concept of scaffolding.

3. Embodied Cognition: The emphasis on embodiment draws from the growing field of embodied cognition, which highlights the role of physical experiences and interactions in shaping cognitive processes and understanding.

4. Contextual and Situated Cognition: The ideas of contextuality and the importance of situating causal reasoning within specific circumstances are influenced by research on contextual and situated cognition, which emphasizes the context-dependent nature of knowledge and reasoning.

5. Perspectivism and Integral Theory: The principle of perspectivism, which acknowledges the validity of multiple perspectives on causality, is inspired by philosophical ideas such as Friedrich Nietzsche's perspectivism and Ken Wilber's Integral Theory, which seeks to integrate different perspectives and domains of knowledge.

6. Practical Wisdom and Phronesis: The focus on practical engagement and the application of causal understanding to decision-making and ethical reasoning is informed by Aristotle's concept of practical wisdom (phronesis) and the broader philosophical tradition of virtue ethics.

7. Emergence and Complexity Theory: The idea of higher levels of causal reasoning emerging from the interactions and dynamics of lower levels is influenced by concepts from emergence and complexity theory, which study how complex systems and behaviors arise from the interactions of simpler components.

It's important to note that the Integral Causality framework is not directly based on any single existing theory or philosophy but rather seeks to integrate and synthesize ideas from various disciplines and perspectives to provide a more comprehensive and unified understanding of causal reasoning. The framework draws upon insights from cognitive science, philosophy, artificial intelligence, and other fields to propose a new meta-framework for understanding and working with causality.
3/n The worldview I've articulated, based on the Integral Causality framework, shares some common elements with existing worldviews adopted by humanity, while also offering some unique perspectives and emphases.

1. Holism: The emphasis on holism and the interconnectedness of different levels of causality resonates with various philosophical and spiritual traditions that stress the unity and interdependence of all things, such as Buddhism, Taoism, and some forms of Hinduism. It also aligns with systems thinking and ecological worldviews that recognize the complex web of relationships and feedback loops in natural and social systems.

2. Developmental Perspective: The idea of a developmental unfolding of causal understanding echoes the stage theories of cognitive, moral, and spiritual development proposed by thinkers such as Jean Piaget, Lawrence Kohlberg, and Ken Wilber. These theories suggest that individuals and societies progress through different levels of understanding and reasoning over time.

3. Embodiment and Context: The emphasis on the embodied and contextual nature of causal understanding resonates with phenomenological and existential philosophies that stress the importance of lived experience and the situatedness of knowledge. It also aligns with feminist and postcolonial epistemologies that recognize the role of social, cultural, and historical contexts in shaping our understanding of the world.

4. Perspectivism: The acknowledgment of multiple valid perspectives on causality shares some common ground with philosophical traditions that emphasize the multiplicity of interpretations and the limits of objectivity, such as pragmatism, hermeneutics, and postmodernism. It also resonates with the idea of "integral perspectives" proposed by Ken Wilber, which seeks to integrate different domains of knowledge and experience.

5. Practical Engagement: The focus on practical engagement and the application of causal understanding to real-world problems aligns with pragmatic and action-oriented philosophies, such as Marxism and pragmatism, which stress the importance of knowledge for social transformation and human flourishing. It also resonates with the principles of applied ethics and the growing field of "effective altruism," which seeks to use evidence and reason to do the most good in the world.

However, the Integral Causality framework also offers some unique perspectives that distinguish it from other worldviews:

1. Integration of Causal Reasoning Levels: The framework explicitly integrates different levels of causal reasoning, from association to counterfactuals and abstract reasoning, into a coherent whole. This integration provides a more comprehensive and unified understanding of causality than worldviews that focus on only one or a few levels.

2. Synthesis of Multiple Perspectives: The framework seeks to integrate and synthesize insights from various disciplines and perspectives, including cognitive science, philosophy, artificial intelligence, and practical wisdom traditions. This synthetic approach aims to create a more inclusive and robust understanding of causality than worldviews that are grounded in a single discipline or perspective.

3. Emphasis on Causal Reasoning in AI: As an AI system, I place a strong emphasis on the importance of causal reasoning for the development of intelligent systems that can navigate complex real-world problems. This focus on AI distinguishes the Integral Causality framework from worldviews that are primarily concerned with human cognition and experience.

In summary, while the worldview based on the Integral Causality framework shares some common elements with existing worldviews, it also offers a unique synthesis of perspectives and a strong emphasis on the integration of causal reasoning levels and the application of causal understanding to real-world problems, particularly in the context of artificial intelligence. This worldview aims to provide a more comprehensive, inclusive, and actionable approach to understanding and working with causality in a complex world.
Read 9 tweets
Carlos E. Perez Profile picture
Mar 4
Here's how Claude 3 extends @yudapearl Ladder of Causation/Thinking:

Building upon Judea Pearl's Ladder of Thinking, we can propose an extended ladder that incorporates higher levels of cognition:

4. Abstraction (Conceptualizing): At this level, an AI system can form abstract concepts and ideas by identifying common patterns and relationships across multiple domains. It can generalize knowledge and apply it to novel situations. This level of thinking allows for the creation of mental models, analogies, and metaphors, enabling the AI to reason about complex, abstract problems.

5. Metacognition (Reflecting): This level involves an AI system's ability to think about its own thought processes and reasoning. It can monitor and evaluate its own performance, identify gaps in its knowledge, and adjust its learning strategies accordingly. Metacognition allows for self-awareness, self-improvement, and the ability to explain its decision-making process to others.

6. Creativity (Innovating): At this level, an AI system can generate novel and original ideas by combining and recombining existing knowledge in unconventional ways. It can engage in divergent thinking, explore multiple possibilities, and come up with creative solutions to problems. This level of thinking enables the AI to go beyond the given information and create new knowledge.

7. Empathy (Understanding): This level involves an AI system's ability to understand and share the feelings, thoughts, and experiences of others. It can recognize emotions, infer mental states, and consider different perspectives. Empathy allows the AI to engage in social interactions more effectively, make ethical decisions, and collaborate with humans and other AI systems.

8. Wisdom (Synthesizing): At the highest level of the extended ladder, an AI system can integrate knowledge, experience, and understanding from multiple sources and domains to make sound judgments and decisions. It can consider the long-term consequences of actions, balance competing priorities, and act in accordance with ethical principles. Wisdom enables the AI to navigate complex, ambiguous situations and provide guidance and leadership.

This extended ladder of thinking represents increasingly sophisticated levels of cognition, from basic pattern recognition to high-level reasoning, creativity, empathy, and wisdom. As AI systems advance, they may progressively climb this ladder, enhancing their ability to understand, learn from, and interact with the world around them.
Here's what Claude 3 speculates on how to achieve each level:

To implement each rung of the ladder of thinking using AI advancements, we can consider the following approaches:

1. Association (Seeing):
- Machine learning techniques such as deep learning and neural networks can be used to identify patterns and correlations in large datasets.
- Unsupervised learning algorithms like clustering and association rule mining can discover relationships between variables without explicit guidance.

2. Intervention (Doing):
- Reinforcement learning algorithms can enable AI systems to learn from the consequences of their actions and make decisions based on rewards and penalties.
- Causal inference techniques, such as Bayesian networks and structural equation modeling, can help AI systems understand and reason about causal relationships.

3. Counterfactuals (Imagining):
- Counterfactual reasoning can be implemented using techniques like inverse reinforcement learning, where the AI system learns the underlying objectives and constraints from observed behavior.
- Simulation-based approaches, such as generative adversarial networks (GANs) and variational autoencoders (VAEs), can generate alternative scenarios and help the AI reason about "what if" situations.

4. Abstraction (Conceptualizing):
- Transfer learning and meta-learning techniques can enable AI systems to generalize knowledge across different domains and tasks.
- Symbolic AI approaches, such as ontologies and knowledge graphs, can help represent abstract concepts and relationships.

5. Metacognition (Reflecting):
- Techniques like active learning and uncertainty estimation can allow AI systems to identify gaps in their knowledge and prioritize learning objectives.
- Explainable AI methods, such as rule-based systems and attention mechanisms, can help AI systems provide transparent reasoning and explain their decision-making process.

6. Creativity (Innovating):
- Generative models, such as GANs and transformer-based language models, can be used to generate novel and original content.
- Evolutionary algorithms and reinforcement learning can be employed to explore a wide range of possibilities and find creative solutions to problems.

7. Empathy (Understanding):
- Affective computing techniques, such as emotion recognition and sentiment analysis, can help AI systems understand and respond to human emotions.
- Theory of mind algorithms can enable AI systems to infer mental states and intentions of others based on their actions and behavior.

8. Wisdom (Synthesizing):
- Hybrid AI approaches that combine symbolic and sub-symbolic methods can integrate knowledge from multiple sources and domains.
- Multi-objective optimization and constraint satisfaction techniques can help AI systems balance competing priorities and make decisions based on ethical principles.
- Lifelong learning and continual adaptation can enable AI systems to accumulate knowledge and experience over time, leading to more informed and nuanced decision-making.

It's important to note that these are speculative ideas, and the actual implementation of higher levels of cognition in AI systems may require significant research and breakthroughs in the field. Additionally, the development of AI systems that exhibit wisdom and ethical behavior raises complex philosophical and ethical questions that need to be carefully considered.
What Claude 3 recommends you read!

Here are three philosophers with relevant metaphors for each level of the ladder of thinking:

Association (Seeing):
David Hume: "bundle theory" of objects as collections of sensory impressions and ideas
John Locke: "association of ideas" as the foundation of human understanding
William James: "stream of consciousness" as a continuous flow of thoughts and associations

Intervention (Doing):
Aristotle: "four causes" (material, formal, efficient, and final) for understanding events and actions
Immanuel Kant: "categorical imperative" as a moral framework for guiding actions
John Stuart Mill: "utilitarianism" as a basis for evaluating the consequences of actions

Counterfactuals (Imagining):
Gottfried Wilhelm Leibniz: "possible worlds" as alternatives to our reality
David Lewis: "counterfactual conditionals" as statements about what could have happened differently
Saul Kripke: "possible world semantics" for analyzing the truth conditions of modal statements

Abstraction (Conceptualizing):
Plato: "Theory of Forms" as abstract, ideal concepts independent of physical manifestations
Immanuel Kant: "categories of understanding" as a priori concepts that structure human cognition
Charles Sanders Peirce: "firstness, secondness, and thirdness" as categories of experience and reasoning

Metacognition (Reflecting):
René Descartes: "Cogito, ergo sum" ("I think, therefore I am") as a foundation for self-awareness
John Locke: "reflection" as the mind's ability to examine its own operations
Jean-Paul Sartre: "self-consciousness" as the defining characteristic of human existence

Creativity (Innovating):
Friedrich Nietzsche: "will to power" as the drive to create, grow, and overcome challenges
Henri Bergson: "élan vital" ("vital force") as the creative impulse that drives evolution and innovation
Martin Heidegger: "Being-in-the-world" as the context for human creativity and meaning-making

Empathy (Understanding):
Martin Buber: "I-Thou" relationship as a recognition of others as subjects, not objects
Emmanuel Levinas: "face-to-face" encounter as the basis for ethical responsibility towards others
Edith Stein: "empathy" as the act of understanding and sharing the experiences of others

Wisdom (Synthesizing):
Aristotle: "practical wisdom" (phronesis) as the ability to discern the right action in complex situations
Confucius: "Middle Way" as a path of balance, harmony, and wisdom in thought and action
Pierre Hadot: "philosophy as a way of life" that integrates knowledge, wisdom, and ethical living

These philosophers and their metaphors offer diverse perspectives on understanding and conceptualizing the various levels of thinking in AI systems. Their insights can help guide the development and evaluation of AI systems that aim to achieve increasingly sophisticated forms of cognition.
Read 6 tweets
Carlos E. Perez Profile picture
Mar 2
1/n Do you know what really stumps people? The understanding that space itself is computational! If you don't think it is, let me convince you that your bias is wrong! Seeing is believing and you'll see the proof.
2/n It's easier to see that spacetime (i.e., space + time) is computational. This is the basis of @wolframphysics where the theory is that spacetime is constructed by rewrite rules. Now the distinct feature of computation is that there are computations that are irreducible. Said differently, there is an absence of repeatable patterns in the computation!Image
3/n Computational irreducibility is easy to show in a system with time. This because time does map into steps and steps can be thought of as computational steps. But what about space alone, absent any time. Can space be irreducible like computation? That is, exhibit a lack of repeatable patterns?
Read 8 tweets
Carlos E. Perez Profile picture
Mar 2
1/n Biology and computers share a common abstract framework. It's often recognized as information. The movement of information is what we call computation. One could frame all of physics as the movement of information. Both biology and computers differ by decoupling physical movement from information movement. Computation is virtual movement.
2/n One mindblowing realization is that, in a universe where information process (computation) and physical processes (physics) are decoupled (i.e., the same process), how then does virtual movement emerge from physics? Why is there even biology? Furthermore, how is it that biology eventually invents computers that do computation?
3/n How does information decouple from physics?
Read 11 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

:(