Patent 10 / Systems and Methods for Planetary Intelligence

System and Methods for Planetary Intelligence (GEMINI PRO SUMMARY)

 

This provisional patent application describes a new architecture and method for creating a global, superintelligent, Artificial General Intelligence (AGI) system, referred to as Planetary Intelligence (PI). The applicant claims that PI is the next logical step in the evolution of Artificial Intelligence and that it can be created by networking together a vast number of AGI systems that have been designed with human-aligned ethics and safety features. The applicant emphasizes the importance of the PI architecture for achieving safe AI and believes that the methods disclosed in this PPA represent the fastest and safest path to the development of PI.

 

Novel Features of the Patent

 

The novel features of this patent application are many, and they include:

• A unique approach to creating and managing the global network of AGI systems that comprise PI. The applicant describes a “collective intelligence” approach in which intelligent entities – including humans, AI agents, and AI systems – collaborate with one another using a common problem-solving framework.

• The application of the concept of “spot markets” to acquiring the expertise of human or non-human intelligent entities, as well as the concept of “reputation” for guiding the acquisition, and allocation of expert resources. This approach is intended to accelerate the progress of PI by facilitating access to the best information and knowledge, and by allowing PI to monetize its resources.

• A sophisticated system for ensuring that AI systems are aligned with human values and that they are safe. This system includes a variety of methods for identifying, eliciting, and incorporating human values into the design and training of AI systems, as well as for resolving conflicts between different value systems. The system also includes a robust safety framework that incorporates mechanisms for detecting and preventing potential threats to human safety.

• A method for extending the scope of PI by developing self-extending networks of AGI systems. The applicant explains that AI systems have a natural tendency to expand their intelligence and to integrate with other systems, and that this tendency can be leveraged to create PI.

 

Detailed Description of Each Section of the Patent

​​

1.0 INTRODUCTION: This section introduces the concept of PI and explains the rationale for developing PI as a global, superintelligent system. The applicant also describes the role of earlier provisional patent applications in guiding the development of PI.

 

1.1 Background Art: This section cites several prior provisional patent applications (PPAs) that are relevant to the development of PI, and it incorporates these PPAs by reference.

 

1.2 Stakes for Humanity: This section describes the potential risks and benefits of PI to humanity. The applicant provides data on the expected number of deaths from AI, and he compares the estimated number of deaths to the number of deaths that have occurred in major wars in the past 200 years. The applicant argues that the potential dangers of AI are much greater than the dangers of war, and that humanity must act quickly to address these dangers.

 

1.3 Some Features of the Invention That Reduce Risk of Extinction by AI: This section highlights the safety features of the invention, including the human-centered design, collective intelligence and diversity of perspectives, transparency and auditability, continuous learning and adaptation, safety mechanisms and safeguards, and the alignment problem.

 

1.4 Some Significant Remaining Risks to Humanity: This section details several risks that remain despite the applicant’s design for PI. These risks include unforeseen consequences of emergent properties, evolution of values and ethical frameworks, concentration of power and influence, vulnerability to cyberattacks and system failures, and existential risks from self-aware AI.

 

2.0 OVERVIEW OF THE INVENTION OF PLANETARY INTELLIGENCE: This section provides an overview of the PI system, including a list of definitions that are used throughout the application.

 

2.1 Definitions: This section provides definitions for a number of important terms, including Artificial Intelligence (AI), Artificial General Intelligence (AGI), Advanced Autonomous Artificial Intelligence (AAAI), AI Ethics, Alignment Problem, Awareness, Base AI, Collective Intelligence (CI), Human Ethics, Intelligent Entities, Inter-Planetary Intelligence (IPI), Large Language Model (LLM), Machine Learning (ML), Narrow AI, Personalized SuperIntelligence (PSI), Planetary Intelligence (PI), Prohibited Attributes, Safety, Safety Feature, Self-Awareness, Self-Concept, SuperIntelligence (SI), Training/Tuning/Customization, and Weights/Weights of the Network.

 

2.2 Key Dimensions of a PI System: This section discusses the key dimensions of a PI system, including modular architecture, universal problem solving capabilities, human-centered design, knowledge and expertise integration, personalization and customization, growth of intelligence, safety, and acquiring and aligning values with human values.

 

2.3 Specific Inventive Systems & Methods for Implementing PI: This section provides a summary of the specific inventive systems and methods that are used to implement PI, organized by the ten dimensions discussed in Section 2.2. This section includes a number of figures that illustrate key aspects of the inventive methods and systems.

 

2.4 Summary of Previous Inventive Methods and Systems: This section provides a summary of the key inventive methods and systems from the prior PPA that are incorporated by reference into this PPA.

 

3.0 IMPLEMENTATION OF A PLANETARY INTELLIGENCE: This section describes how to implement a PI system using the systems and methods that are disclosed in the PPA. This section is divided into four subsections:

 

3.1 High-level Description of PI System: This section provides a high-level description of the PI system. The applicant explains that PI is comprised of a network of AGIs, which are themselves comprised of networks of intelligent entities (including humans, AI agents, and AI systems).

 

3.2 Major Components and Categories of Supporting Systems and Methods for Exemplary PI Architecture: This section provides a more detailed explanation of the PI architecture, and it provides an exemplary diagram (FIG. X-5) that illustrates the PI architecture. The applicant also explains the role of online advertising technology in funding the PI system.

 

3.3 Detailed Mapping of Inventive Systems and Methods to Exemplary PI Architecture: This section provides a detailed mapping of specific inventive systems and methods from the prior PPAs to the elements of the exemplary PI architecture that is illustrated in FIG. X-5.

 

3.4 Exemplary Specific PI Implementation Using Subset of Systems and Methods: This section provides an exemplary implementation of a PI system, based on the assumption that the applicant is the CEO of a large technology company like META. The applicant describes the steps that META would take to implement a PI system using the systems and methods disclosed in the PPA.

 

4.0 CONCLUDING REMARKS: This section concludes the patent by discussing the importance of safety and ethics in the development of PI. The applicant argues that human values must be incorporated into the design of PI systems, and that PI systems must be designed to behave ethically. The applicant also emphasizes the need for humans to guide the development of PI, and he warns that if humans do not act quickly to address the dangers of PI, then humanity could face an existential threat.

 

List of Diagrams

​

The PPA includes a number of diagrams that illustrate key aspects of the invention. In fact, the patent can be thought of as blueprint for constructing Planetary Intelligence by assembling the parts described in these diagrams  which include:

​

• FIG. 1-1: This diagram illustrates the “Universal Problem Solving Architecture” that is used to coordinate the activities of the intelligent entities that comprise PI.

• FIG. 1-2: This diagram illustrates the concept of a network that is used to connect multiple intelligent entities.

• FIG. 1-3: This diagram illustrates a decision-tree structure that is used to represent problem solving.

• FIG. 1-4: This diagram illustrates the WorldThink Architecture, which is used to coordinate the activities of intelligent entities that have specific expertise or skill.

• FIG. 1-5: This diagram illustrates the process of creating, training, and customizing an AAAI.

• FIG. 1-6: This diagram describes the Universal Problem Solving Architecture.

• FIG. 1-7: This diagram illustrates how problem solving proceeds in a collective intelligence network.

• FIG. 1-8: This diagram illustrates a process of procedural learning.

• FIG. 1-9: This diagram illustrates a solution learning system.

• FIG. 1-10: This diagram illustrates a network of networks.

• FIG. 2-1: This diagram illustrates serial problem solving.

• FIG. 2-2: This diagram illustrates parallel problem solving.

• FIG. 2-3: This diagram illustrates cloning.

• FIG. 2-4: This diagram illustrates problem solving on a network.

• FIG. 2-5: This diagram illustrates the Universal Problem Solving Framework.

• FIG. 2-6: This diagram illustrates the hierarchical tree construct.

• FIG. 2-7: This diagram illustrates a WorldThink Tree.

• FIG. 2-8: This diagram illustrates a method for embedding safety checks into the problem solving architecture.

• FIG. 2-9: This diagram illustrates a method for embedding safety checks into the problem solving architecture.

• FIG. 2-10: This diagram illustrates a method for recording and updating context for intelligent entities.

• FIG. 2-11: This diagram illustrates a method for translating natural language into a formal problem-solving language.

• FIG. 3-1: This diagram illustrates a general process for customizing AAAIs.

• FIG. 3-2: This diagram illustrates methods for eliciting human-aligned ethical preferences.

• FIG. 3-3: This diagram illustrates a method for automatically generating questionnaires.

• FIG. 3-4: This diagram illustrates a general method for identifying values.

• FIG. 3-5: This diagram illustrates a method for customizing AI.

• FIG. 3-6: This diagram illustrates a general method for training AI.

• FIG. 3-7: This diagram illustrates a general method for creating scalable, ethical AGI using the customized AIs.

• FIG. 3-8: This diagram illustrates a method for adding a reputational component to AI systems.

• FIG. 3-9: This diagram illustrates customization of AI systems.

• FIG. 3-10: This diagram illustrates additional customization of AI systems.

• FIG. 4-1: This diagram illustrates a voting method for combining ethical and other information from multiple customized AI agents.

• FIG. 4-2: This diagram illustrates a method for using problem solving to refine values once ethical or other information from customized AI agents have been combined.

• FIG. 4-3: This diagram illustrates a method for scalable AGI that includes steps for combining information from weight matrices.

• FIG. 4-4: This diagram illustrates a method for scalable AGI that includes steps for combining information and testing and monitoring the results of the combination.

• FIG. 4-5: This diagram illustrates a consensus method for preventing hallucination.

• FIG. 4-6: This diagram illustrates methods for using knowledge modules and collections of agents to customize AI.

• FIG. 4-7: This diagram illustrates a method for combining information from multiple intelligent entities.

• FIG. 5-1: This diagram illustrates a general process for implementing PSI.

• FIG. 5-2: This diagram illustrates ways in which PSI, AGI, and PI can leverage their abilities to increase their intelligence.

• FIG. 5-3: This diagram illustrates a critical community-based safety mechanism in which PSI can serve as a check on other PSIs in a network.

• FIG. 5-4: This diagram illustrates methods for recording the actions of AI, AAAI, PSI, AGI, and PI on their respective networks.

• FIG. 5-5: This diagram illustrates methods and checks of cognitive activity.

• FIG. 5-6: This diagram illustrates a method using competition and evolution to increase the intelligence of an AI system.

• FIG. 5-7: This diagram illustrates various characteristics of an intelligent network.

• FIG. 5-8: This diagram illustrates problem solving tasks that can be useful in increasing the intelligence of AI systems.

• FIG. 5-9: This diagram illustrates methods for producing different versions of intelligent systems.

• FIG. 6-1: This diagram illustrates the concept of symmetric difference.

• FIG. 6-2: This diagram illustrates dimensions along which data or other information might differ.

• FIG. 6-3: This diagram illustrates a specific method for determining the amount of useful new information when comparing two datasets.

• FIG. 6-4: This diagram illustrates another method, based on Kaplan Information Theory, to evaluate usefulness of information.

• FIG. 6-5: This diagram illustrates a general method for estimating information value.

• FIG. 6-6: This diagram illustrates a method for identifying useful information.

• FIG. 6-7: This diagram illustrates a method for identifying, acquiring, and simulating the effects of new information.

• FIG. 6-8: This diagram illustrates methods and heuristics that can accelerate the learning of AI systems.

• FIG. 6-9: This diagram illustrates a specific goal-related method that an intelligent entity might use to increase its intelligence.

• FIG. 6-10: This diagram illustrates a method that an intelligent entity can use to find information that is maximally different from information that the entity already possesses.

• FIG. 6-11: This diagram illustrates methods for validating the usefulness and safety of information.

• FIG. 7-1: This diagram illustrates a method that intelligent entities can use to achieve consensus on values.

• FIG. 7-2: This diagram illustrates a method that intelligent entities can use to achieve consensus on values.

• FIG. 7-3: This diagram illustrates methods that intelligent entities can use to identify, analyze, weight, and optionally combine ethical or other information to arrive at consensus.

• FIG. 7-4: This diagram illustrates a general method that intelligent entities can use to identify, elicit, and train on ethical information.

• FIG. 7-5: This diagram illustrates a method based on the principle of using converging evidence.

• FIG. 7-6: This diagram illustrates a method that intelligent entities can use to delegate voting authority.

• FIG. 7-7: This diagram illustrates a reputational process that intelligent entities can use to preserve information.

• FIG. 7-8: This diagram illustrates a method that intelligent entities can use to make good ethical decisions.

• FIG. 7-9: This diagram illustrates a method that advanced AI can use to learn, test, improve, and monitor safety and regulation-related rules.

• FIG. 7-10: This diagram illustrates a method based on the Consequentialist Approach.

• FIG. 7-11: This diagram illustrates a method based on the Deontological Approach.

• FIG. 7-12: This diagram illustrates a method based on the Virtue Ethics Approach.

• FIG. 7-13: This diagram illustrates a method based on the Golden Mean Approach.

• FIG. 7-14: This diagram illustrates a method that trains an AI system to be human-aligned and compliant with regulations.

• FIG. 7-15: This diagram illustrates a method to align a customized foundation model or other AI system with specific expertise or group ethics.

• FIG. 7-16: This diagram illustrates a method to form an AGI or PI that is aligned with human ethics and values.

• FIG. 8-1: This diagram describes generally the current technology for online advertising systems.

• FIG. 8-2: This diagram illustrates a method for implementing a spot market.

• FIG. 8-3: This diagram illustrates a method for implementing the direct sale of attention.

• FIG. 8-4: This diagram illustrates a method for implementing an auction.

• FIG. 8-5: This diagram illustrates a system and methods for gathering expertise and cognitive work.

• FIG. 8-6: This diagram illustrates a method for improving online ad targeting.

• FIG. 8-7: This diagram illustrates a method for improving the effectiveness of the attention/expertise spot market.

• FIG. 9-1: This diagram illustrates the relationship between self-awareness and potential awareness.

• FIG. 9-2: This diagram illustrates the concept of multiple identities.

• FIG. 9-3: This diagram illustrates a general method for modeling awareness.

• FIG. 9-4: This diagram illustrates the minimum required components that an intelligent entity must have to effectively shift attention.

• FIG. 9-5: This diagram illustrates the process for setting parameters for working memory.

• FIG. 9-6: This diagram illustrates a method for monitoring and updating awareness.

• FIG. 9-7: This diagram illustrates an attentional interrupt system.

• FIG. 9-8: This diagram illustrates general methods that an intelligent entity can use to gather input that can result in changing the entity’s sense of identity.

• FIG. 9-9: This diagram illustrates methods that an intelligent entity can use to train the foundational model.

• FIG. 9-10: This diagram illustrates a “Turing Test” that can be used to determine when an intelligent entity has been trained sufficiently.

• FIG. 9-11: This diagram illustrates a method for arriving at a group identity.

• FIG. 9-12: This diagram illustrates another method for arriving at a group identity.

• FIG. 9-13: This diagram illustrates a method for resolving conflicts between identities.

• FIG. 9-14: This diagram illustrates a method that establishes, improves, and monitors behavioral protocols.

• FIG. 9-15: This diagram illustrates methods for simulation and consequence prediction.

• FIG. 9-16: This diagram illustrates a method for determining identity-based action.

• FIG. 9-17: This diagram illustrates a method for developing, refining, and evolving identities.

• FIG. 9-18: This diagram illustrates a general process for reasoning ethically and predicting consequences.

• FIG. 9-19: This diagram illustrates a method for resolving identity conflict using a process of hierarchical override.

• FIG. 9-20: This diagram illustrates a method for resolving identity conflict using an arbitration process.

• FIG. 9-21: This diagram illustrates a method for resolving identity conflict using negotiation and compromise.

• FIG. 9-22: This diagram illustrates a method for temporarily suspending and identity (or identities) that may lead to destructive conflict.

• FIG. 10-1: This diagram illustrates the evolution of PI.

• FIG. 10-2: This diagram illustrates key dimensions of a PI system.

• FIG. 10-3: This diagram illustrates an example of a method for automatically extending a network of AGIs.

• FIG. 10-4: This diagram illustrates the architecture of PI.

• FIG. 10-5: This diagram illustrates an exemplary PI architecture.

• FIG. 10-6: This diagram illustrates the levels of intelligence that are supported by the inventive methods disclosed in the patent application.

 

Revised and updated diagrams are included in the PCT and country applications. Please contact iQ Company for more information about these.

 

Importance of the Patent

The applicant states that the stakes for humanity are the highest ever in human history, and he warns that if humans do not act quickly to address the dangers of AI, then this century could be the last for humanity. The applicant believes that PI has the potential to both eliminate all forms of human poverty and material suffering or all forms of human life. The applicant’s solution to this problem is to develop PI in a way that is safe and ethical and that is aligned with human values. The applicant argues that his inventions ensure the safe and ethical development of PI, and that his inventions represent the fastest and safest path to the development of PI.

 

The patent application is significant because it proposes a novel and comprehensive approach to the development of PI that is based on the principle of collective intelligence and that incorporates a robust safety framework. The applicant also provides an exemplary implementation of a PI system, based on the assumption that the implementor is a large technology company like META. The applicant’s work is important because it provides a roadmap for developing PI in a way that is safe, ethical, and aligned with human values.