# ASI and RSI Timeline Ontology **Entity class:** Research ontology / navigation index **Domain:** Artificial intelligence / Scientific discovery / Biotechnology / Neuroscience / Compute / Governance / Physical systems **Maturity:** Developed ## Purpose This ontology converts the entities, programs, companies, software systems, scientific concepts, and strategic mechanisms in [[research/ASI and RSI Timeline Research Moonshots|ASI and RSI Timeline Research Moonshots]] into independently navigable wiki objects. It is designed as the exploration layer that precedes Reminder promotion: the source can now route outward into developed concepts, and every concept routes back to its evidence-bearing research dossier and field hubs. ## Canonicalization notes - The broadcast's apparent **CCI** reference is routed to [[CZI Virtual Cells Platform|CZI]], the Chan Zuckerberg Initiative program supported by the primary organizational source. - **Leila Science** is normalized to [[Lila Sciences]]. - **Jean-Marie King** is normalized to [[Jean-Rémi King]]. - **Tavis** is normalized to [[Tavus]]. - **Tower Therapeutics** is routed to [[Tahoe Therapeutics]] where the experimental context and primary company evidence agree. - Existing canonical pages are retained for [[Functional Magnetic Resonance Imaging|fMRI]], [[Cellular Digital Twins|digital twins of cells]], [[Digital Twin]], [[Superintelligence|artificial superintelligence]], [[Open-Ended Evolution|open-ended algorithms]], and [[neural foundation models|neural foundation models]]. ## Scientific Intelligence and Recursive Improvement Scientific discovery loops, recursive improvement, open-ended search, and machine-generated scientific formalisms. **Field routes:** [[AI for Science]] · [[Scientific Acceleration]] · [[Machine Intelligence]] - [[Richard Socher]] — Richard Socher is an artificial-intelligence researcher, entrepreneur, and author whose work spans natural-language processing, search, AI for science, and recursive self-improvement. - [[The Eureka Machine]] — The Eureka Machine is Richard Socher's framework for using artificial intelligence across the scientific process, from ideation and hypothesis generation through experiment, measurement, and validation. - [[Recursive]] — Recursive is the AI company discussed in the broadcast as Richard Socher's effort to develop systems capable of improving components of their own research and development process. - [[Recursive Self-Improvement]] — Recursive self-improvement is the process by which an intelligent system contributes to improving the models, tools, training procedures, evaluations, or computational systems used to produce its successors. - [[Weak RSI]] — Weak recursive self-improvement describes bounded cases in which AI improves selected components of an AI-development process without autonomously controlling the complete cycle. - [[Strong RSI]] — Strong recursive self-improvement describes a more complete loop in which an AI system can redesign and validate consequential parts of its own architecture, training, tools, and successor systems. - [[Scientific Superintelligence]] — Scientific superintelligence is machine intelligence that exceeds human scientific capability across hypothesis formation, formal reasoning, experimental design, instrumentation, interpretation, and discovery. - [[Full-Stack AI for Science]] — Full-stack AI for science integrates artificial intelligence across the complete scientific workflow: question formation, hypothesis generation, experimental implementation, measurement, analysis, theory revision, and replication. - [[Automated Scientific Method]] — An automated scientific method is a machine-operated loop that proposes hypotheses, selects experiments, executes or delegates them, measures outcomes, and updates its model from the results. - [[Hypothesis-to-Experiment Compression]] — Hypothesis-to-experiment compression is the reduction of time and coordination required to move from a scientific conjecture to an empirical test and interpretable result. - [[Ideation-Implementation-Validation Loop]] — The ideation-implementation-validation loop models science as repeated movement from a candidate idea, through a physical or computational test, to evidence that changes the next idea. - [[Open-Endedness]] — Open-endedness is the capacity of a search or learning process to continue producing novel structures, behaviors, questions, or capabilities rather than converging quickly on one fixed objective. - [[Evolutionary Search]] — Evolutionary search explores a design space by generating variation, evaluating candidates, preserving useful changes, and recombining or mutating successful structures. - [[Agent Swarm]] — An agent swarm is a coordinated population of AI agents that explores multiple subproblems, hypotheses, tools, or solution paths in parallel. - [[Automated Hypothesis Generation]] — Automated hypothesis generation uses models to propose causal explanations, experimental questions, or candidate mechanisms from data and prior knowledge. - [[Automated Experimentation]] — Automated experimentation uses software, robotics, and instrumentation to execute experimental protocols with machine-controlled scheduling, measurement, and adaptation. - [[Robotic Laboratory]] — A robotic laboratory is an instrumented experimental environment in which machines prepare samples, operate equipment, capture measurements, and execute protocols under software control. - [[Scientific Domain Collapse]] — Scientific domain collapse is the compression of organizational and temporal boundaries that ordinarily separate disciplines, hypotheses, experiments, analysis, and publication. - [[Abstraction Layers in Science]] — Abstraction layers in science allow researchers to work with higher-level objects and relationships without manually reproducing every lower-level calculation or laboratory operation. - [[Science in Natural Language]] — Science in natural language is the use of conversational language as an interface to formal models, databases, instruments, simulations, and experimental workflows. - [[Machine-Generated Scientific Formalism]] — Machine-generated scientific formalism is a mathematical, symbolic, or executable representation proposed by an AI system to organize observations and make testable predictions. - [[Simulation-Verification Boundary]] — The simulation-verification boundary separates domains where candidate solutions can be cheaply modeled or checked from domains where feedback remains slow, destructive, ambiguous, or inaccessible. - [[Computational-Substrate Self-Modification]] — Computational-substrate self-modification is the ability of machine intelligence to influence or redesign the hardware, fabrication process, runtime, or physical systems on which it depends. - [[The Bitter Lesson]] — The Bitter Lesson is Richard Sutton's observation that general methods able to exploit increasing computation have repeatedly outperformed systems built primarily from human domain knowledge. - [[General Function Approximation]] — General function approximation treats learning systems as mechanisms for estimating broad classes of relationships from data rather than as collections of hand-authored rules for one task. - [[Fixed Point of Recursive Self-Improvement]] — A fixed point of recursive self-improvement is a hypothetical state at which a system can no longer produce meaningful improvements to its own architecture, tools, or research process under the resources and constraints available to it. - [[Metacognition]] — Metacognition is the capacity to monitor, evaluate, and regulate one's own cognitive processes, including confidence, error detection, strategy selection, and awareness of uncertainty. - [[Simulatable Domain]] — A simulatable domain is a field in which relevant states, interactions, or outcomes can be represented computationally well enough to support useful prediction, experimentation, or design. ## Computational Biology and Virtual Cells Virtual cells, organoids, biological foundation models, programmable organisms, and the model-to-laboratory loop. **Field routes:** [[AI-Driven Biology]] · [[Cellular Digital Twins]] · [[Computational Biology]] - [[Virtual Cell]] — A virtual cell is a data-driven or mechanistic computational model intended to predict cellular identity, state, and response under genetic, chemical, environmental, or disease perturbations. - [[Cell Simulator]] — A cell simulator is a computational system that predicts how a cell changes over time or responds to a specified intervention. - [[Whole-Cell Model]] — A whole-cell model attempts to represent many interacting cellular processes within one executable framework rather than modeling a single pathway in isolation. - [[CZI Virtual Cells Platform]] — The CZI Virtual Cells Platform is the Chan Zuckerberg Initiative's platform for sharing cellular datasets, models, benchmarks, and tools that support development of AI-based virtual-cell systems. - [[Tahoe Therapeutics]] — Tahoe Therapeutics develops large-scale perturbational single-cell data and models intended to support virtual-cell research and drug discovery. - [[Massive Cellular Perturbation Study]] — A massive cellular perturbation study applies many genetic, chemical, or environmental interventions across large numbers of cells and measures the resulting state changes. - [[Single-Cell Perturbation Data]] — Single-cell perturbation data records how individual cells respond to controlled interventions, commonly through transcriptomic, proteomic, imaging, or multimodal measurements. - [[Digital Biology]] — Digital biology treats biological measurements, models, simulations, and programmable interventions as an integrated computational discipline. - [[Bioinformatics]] — Bioinformatics develops computational methods for storing, integrating, analyzing, and interpreting biological data such as sequences, structures, expression profiles, and clinical annotations. - [[Programmable Biology]] — Programmable biology is the design of biological systems through controllable genetic, cellular, molecular, or ecological interventions. - [[Biology as an Engineering Discipline]] — Biology becomes an engineering discipline when living systems can be measured, modeled, designed, built, tested, and iteratively improved against explicit objectives. - [[Biological Simulation]] — Biological simulation represents molecular, cellular, organismal, or ecological processes computationally so that hypotheses and interventions can be explored before physical testing. - [[Personalized Cellular Simulation]] — Personalized cellular simulation uses measurements from a particular person or patient-derived cells to predict individual biological responses. - [[Patient-Specific Cell Model]] — A patient-specific cell model is calibrated to biological material or measurements from one person rather than only to a population average. - [[Multicellular Organoid Network]] — A multicellular organoid network links multiple tissue-like organoids so that signaling, immune interaction, metabolism, and treatment effects can be studied across compartments. - [[Whole Lymphatic-System Organoid]] — A whole lymphatic-system organoid is a proposed connected set of lymphatic and immune tissue models intended to reproduce interactions that a single lymph-node organoid cannot capture. - [[Lymph-Node Organoid]] — A lymph-node organoid is a three-dimensional human-cell model designed to reproduce selected structures and immune functions of lymph-node tissue. - [[Pluripotent Stem Cell]] — A pluripotent stem cell can differentiate into cell types derived from all three embryonic germ layers while continuing to self-renew under suitable conditions. - [[Patient-Derived Organoid]] — A patient-derived organoid is grown from a person's cells to preserve selected genetic and phenotypic features of that person's tissue or disease. - [[Animal-Free Toxicity Testing]] — Animal-free toxicity testing evaluates harmful biological effects through human cells, organoids, organs-on-chips, computational models, or other non-animal methods. - [[Parallel Bio]] — Parallel Bio develops human immune-system models, including lymph-node organoids, for drug, vaccine, and toxicity testing. - [[Lila Sciences]] — Lila Sciences develops scientific-superintelligence systems and AI-operated laboratories for life, chemical, and materials science. - [[Periodic Labs]] — Periodic Labs develops AI scientists and autonomous laboratories for physical-science discovery. - [[Wetware Experimentation]] — Wetware experimentation is physical experimentation on living cells, tissues, organisms, or biological materials rather than purely computational evaluation. - [[Protein-State Measurement]] — Protein-state measurement captures the identity, abundance, modification, localization, or interaction state of proteins within a biological sample. - [[Nondestructive Cellular Measurement]] — Nondestructive cellular measurement observes biologically meaningful state while keeping the same cell alive for later observation. - [[Individualized Drug Simulation]] — Individualized drug simulation predicts how a proposed treatment may affect a particular person's modeled cells, tissues, or physiological state. - [[Biological Foundation Model]] — A biological foundation model is a broadly trained model intended to represent reusable structure across biological sequences, cells, molecules, images, or experimental measurements. - [[Organoid System]] — An organoid system uses self-organizing three-dimensional cell cultures to reproduce selected structures or functions of tissues and organs. - [[Immunotherapy Testing]] — Immunotherapy testing evaluates whether treatments can activate, redirect, or restore immune responses against cancer or other disease targets. - [[Proxima Labs]] — Proxima Labs is discussed in the Moonshots research source as part of the emerging ecosystem of AI-assisted scientific and biological experimentation. - [[Ignota Labs]] — Ignota Labs is discussed in the Moonshots research source as a company applying machine intelligence to scientific and therapeutic discovery. - [[Jeff von Maltzahn]] — Jeff von Maltzahn is a biotechnology entrepreneur associated with AI-enabled scientific companies and platforms discussed in the Moonshots research source. - [[Robotic Scientific Facility]] — A robotic scientific facility integrates automated instruments, sample handling, measurement, and software control into a repeatable experimental environment. - [[DNA-Triggered Behavioral Program]] — A DNA-triggered behavioral program is an inherited developmental mechanism through which genetic regulation helps construct neural and bodily systems predisposed toward species-typical behavior. - [[Gene Drive]] — A gene drive biases inheritance so that a genetic trait can spread through a population more rapidly than ordinary Mendelian transmission would predict. - [[Kevin Esvelt]] — Kevin Esvelt is a biological engineer known for work on CRISPR-based gene drives and for advocating transparent, community-guided approaches to technologies that can alter wild populations. - [[Genetically Propagated Mosquito Sterility]] — Genetically propagated mosquito sterility uses inherited biological mechanisms to reduce reproduction in disease-carrying mosquito populations. - [[Engineered Inheritance]] — Engineered inheritance deliberately changes the probability that a designed genetic trait passes from one generation to the next. - [[Population-Scale Genetic Intervention]] — A population-scale genetic intervention seeks to alter the prevalence or characteristics of a trait across a reproducing population rather than treating organisms one at a time. - [[Sterile Insect Technique]] — The sterile insect technique suppresses pest or disease-vector populations by releasing large numbers of sterile males whose matings produce no viable offspring. - [[Wolbachia Vector Control]] — Wolbachia vector control introduces or amplifies strains of the bacterium Wolbachia in mosquito populations to reduce pathogen transmission or reproductive success. - [[Dengue Suppression]] — Dengue suppression combines surveillance, mosquito control, vaccination where appropriate, and biological interventions to reduce transmission of dengue viruses. - [[Zika Suppression]] — Zika suppression aims to reduce transmission of Zika virus through vector control, reproductive-health measures, surveillance, and outbreak response. - [[Yellow Fever Suppression]] — Yellow fever suppression uses vaccination, mosquito control, surveillance, and rapid response to prevent urban and sylvatic transmission from producing epidemics. - [[Malaria Control]] — Malaria control combines prevention, diagnosis, treatment, vector management, vaccines, and surveillance to reduce illness, death, and transmission caused by Plasmodium parasites. - [[Tick Suppression]] — Tick suppression reduces populations or pathogen transmission through habitat management, host treatment, biological controls, pesticides, vaccines, or genetic intervention. - [[Invasive-Species Engineering]] — Invasive-species engineering uses genetic, reproductive, behavioral, or ecological technologies to suppress or alter organisms that damage an ecosystem outside their native range. - [[Biological Control System]] — A biological control system uses living organisms, inherited traits, pathogens, or ecological interactions to regulate another population. - [[Programmable Organism]] — A programmable organism is a living system whose sensing, regulation, metabolism, or behavior has been deliberately engineered to perform specified functions. - [[Programming Biology]] — Programming biology treats DNA, gene regulation, cells, and organisms as substrates whose behavior can be predictably redesigned. - [[Ecosystem Genetic Engineering]] — Ecosystem genetic engineering intentionally changes inherited traits in wild or managed populations to alter ecological outcomes. - [[Colossal Biosciences]] — Colossal Biosciences is a biotechnology company developing genetic and reproductive technologies around de-extinction, species restoration, and conservation. - [[De-Extinction Technology]] — De-extinction technology uses ancient DNA, genome engineering, selective breeding, cloning, and reproductive biology to create organisms resembling extinct species or restoring selected lost traits. - [[Federal Biotechnology Deployment]] — Federal biotechnology deployment is the coordinated use of public funding, regulation, procurement, infrastructure, and partnerships to move biological technologies from research into national capability. - [[Synthetic Ecology]] — Synthetic ecology designs or steers communities of organisms and their interactions to perform ecological functions. - [[Biotechnology Panic]] — Biotechnology panic is a public or political reaction in which legitimate biological risks are generalized into opposition to an entire field or capability. - [[Biological-World Engineering]] — Biological-world engineering treats organisms, populations, and ecosystems as increasingly designable systems while recognizing that living systems reproduce, evolve, and escape laboratory boundaries. ## Neural Decoding and Cognitive Systems Brain imaging, neural encoding and decoding, cognition, biological algorithms, and high-bandwidth brain-computer interfaces. **Field routes:** [[Neural Decoding]] · [[Brain-Computer Interfaces]] · [[Functional Magnetic Resonance Imaging]] - [[Brain-IT]] — Brain-IT is an AI model developed in Michal Irani's lab at the Weizmann Institute of Science to reconstruct viewed images from fMRI activity and improve learning through a reverse image-to-brain encoding pathway. - [[Michal Irani]] — Michal Irani is a computer-vision researcher at the Weizmann Institute of Science whose laboratory developed Brain-IT. - [[Neural Encoding Model]] — A neural encoding model predicts measured neural activity from a stimulus, task, or environmental variable. - [[Reverse Brain Encoder]] — A reverse brain encoder predicts how measured brain regions would respond to an input image or other stimulus. - [[Synthetic fMRI Training Data]] — Synthetic fMRI training data consists of predicted brain-imaging responses generated by an encoding model rather than acquired directly in a scanner. - [[Self-Generating Brain-Imaging Dataset]] — A self-generating brain-imaging dataset is produced when a model uses one learned direction, such as image-to-fMRI encoding, to manufacture additional training pairs for the reverse decoding task. - [[Visual-Cortex Reconstruction]] — Visual-cortex reconstruction estimates an image or visual representation from neural activity associated with seeing, imagining, or recalling visual content. - [[Perception Decoding]] — Perception decoding infers what a person is currently seeing, hearing, or otherwise sensing from measured neural activity. - [[Language Decoding]] — Language decoding maps neural signals to words, phonemes, semantic representations, or intended speech. - [[Imagined-Image Decoding]] — Imagined-image decoding attempts to reconstruct visual content a person generates internally without a corresponding image present in the environment. - [[Dream Decoding]] — Dream decoding uses neural and physiological measurements to infer categories, imagery, or narrative features associated with dreaming. - [[Dream Reconstruction]] — Dream reconstruction is the generation of images, language, or other media intended to approximate dream content from recorded brain activity and reported experience. - [[Hallucination Reconstruction]] — Hallucination reconstruction attempts to infer or visualize internally generated perceptual content from neural measurements. - [[Brain-State Decoding]] — Brain-state decoding infers a cognitive, perceptual, affective, or task-related state from neural measurements. - [[Person-Specific Neural Decoder]] — A person-specific neural decoder is trained or calibrated for one individual's neural anatomy and signal patterns. - [[Gallant Lab]] — The Gallant Lab at the University of California, Berkeley studies how sensory and semantic information are represented in the brain and develops computational encoding and decoding models. - [[Jack Gallant]] — Jack Gallant is a neuroscientist at the University of California, Berkeley known for research on computational models of sensory and semantic representation in the human brain. - [[Jean-Rémi King]] — Jean-Rémi King is a researcher whose work connects artificial intelligence with neural decoding of language and perception. - [[Meta Neural Decoding Research]] — Meta neural-decoding research includes work on mapping neural recordings to language, perception, and model-derived representations. - [[Neuralink Electrophysiological Data]] — Neuralink electrophysiological data consists of high-temporal-resolution neural recordings produced through implanted electrode systems. - [[Scaling Laws for Neural Foundation Models]] — Scaling laws for neural foundation models study how decoding or representation performance changes with more participants, recording time, channels, model capacity, and compute. - [[Full-Bandwidth Brain-Computer Interface]] — A full-bandwidth brain-computer interface is a proposed system capable of reading and writing a sufficiently rich range of neural signals to support fluid, high-dimensional interaction. - [[Full-Dive Virtual Reality]] — Full-dive virtual reality is an immersive interface in which sensory experience and action are mediated directly enough to create the functional experience of entering a synthetic environment. - [[Direct Cognitive Communication]] — Direct cognitive communication is the transmission of intended meaning or internal representations through neural interfaces without relying entirely on speech, typing, or conventional gesture. - [[Thought-Perception Distinction]] — The thought-perception distinction separates cognition generated internally from neural processing tied to an external sensory event. - [[Thought Cloud Cognition]] — Thought-cloud cognition describes thinking experienced as spatial, imagistic, relational, affective, or simultaneous rather than as an internal sequence of sentences. - [[Text-Free Neural Model]] — A text-free neural model is trained without language tokens as its primary representational substrate, relying instead on images, actions, sensory streams, or other structured signals. - [[Image-Native Cognition]] — Image-native cognition treats visual and spatial representations as primary structures of thought rather than as material translated from language. - [[Biological Learning Algorithm]] — A biological learning algorithm is a hypothesized computational description of how nervous systems update behavior and internal representations using local biological processes. - [[C. elegans Whole-Brain Simulation]] — A C. elegans whole-brain simulation models the nervous system of the nematode whose compact connectome makes organism-scale neural analysis unusually tractable. - [[Liquid AI]] — Liquid AI develops neural architectures influenced by continuous-time and dynamical-system approaches associated with liquid neural networks. - [[Distributed Biological Memory]] — Distributed biological memory is the hypothesis that learned state can depend on bodily, molecular, electrical, or developmental organization beyond a conventional brain-centered account. - [[Regenerated-Brain Memory Persistence]] — Regenerated-brain memory persistence refers to reported cases in which an organism shows previously learned behavior after losing and regrowing neural tissue. - [[Voxel-Level Brain Imaging]] — Voxel-level brain imaging analyzes activity at the spatial units used in volumetric neuroimaging rather than treating an entire region as one undifferentiated signal. - [[High-Temporal-Resolution Neural Data]] — High-temporal-resolution neural data records changing brain activity quickly enough to resolve the sequence and timing of cognitive events. - [[Internal Monologue]] — Internal monologue is the experience of thought in language-like form without overt speech. - [[Sentence Thinker]] — A sentence thinker habitually experiences a substantial portion of conscious thought in words, phrases, or internally articulated sentences. - [[Nonverbal Thinker]] — A nonverbal thinker relies substantially on imagery, spatial relations, affect, bodily simulation, or abstract structure rather than continuous inner speech. - [[Emoji as Emotional Symbol]] — An emoji is a compact visual symbol that can carry affect, stance, irony, social intent, or relational nuance that words alone may underspecify. - [[Reverse-Engineered Nervous System]] — A reverse-engineered nervous system is a computational account that attempts to reproduce the functional organization linking an organism's sensory input, internal dynamics, and behavior. - [[Biological Neural Dynamics]] — Biological neural dynamics are the time-dependent patterns produced by interacting neurons, circuits, bodies, and environments. - [[Sam Gershman]] — Sam Gershman is a computational cognitive scientist whose work examines learning, inference, representation, and the algorithms underlying human and animal behavior. - [[Latent Behavioral Algorithm]] — A latent behavioral algorithm is an inferred internal rule or computation that explains how an organism converts perception, state, and memory into action. ## Generative Humans, Continuity, and Presence Interactive avatars, synthetic presence, preserved biological materials, personality reconstruction, and identity continuity. **Field routes:** [[Digital Human Twin]] · [[Machine Succession]] · [[Posthumous Avatar]] - [[Tavus]] — Tavus develops real-time conversational video systems and human-interaction models designed to perceive and generate face-to-face behavior. - [[Griffin]] — Griffin is Tavus's Human Interaction Model for real-time, full-duplex video conversation. - [[Video Turing Test]] — A video Turing test evaluates whether people interacting through live audiovisual communication can reliably distinguish an AI-generated participant from a human participant. - [[Human Interaction Model]] — A Human Interaction Model is Tavus's term for a model that jointly perceives and generates real-time audiovisual social behavior rather than chaining separate turn-taking components. - [[Full-Duplex AI Conversation]] — Full-duplex AI conversation allows a system to listen, perceive, reason, react, and speak while another participant is still acting or talking. - [[NVIDIA Video Full-Duplex Benchmark]] — The NVIDIA Video Full-Duplex Benchmark evaluates perception, generation, and conversational timing in real-time audiovisual AI systems. - [[Synthetic Presence]] — Synthetic presence is the experience of an artificial participant appearing socially present through coordinated voice, timing, attention, expression, embodiment, and responsiveness. - [[Real-Time Interactive Avatar]] — A real-time interactive avatar is a generated audiovisual persona that responds to a user and environment with sufficiently low latency for live conversation. - [[Generative Human]] — A generative human is an artificial audiovisual person produced dynamically by a model rather than replayed from a fixed recording or animated through a limited script. - [[End-to-End Generative Pixels]] — End-to-end generative pixels describes a system that generates the visible video scene directly rather than compositing a fixed avatar from separately controlled parts. - [[Interactive Generative Video]] — Interactive generative video creates visual scenes that change in real time in response to users, events, or model decisions. - [[Wanstreamer]] — Wanstreamer is discussed in the broadcast as an open model demonstrating low-latency interactive generative video from the Alibaba ecosystem. - [[Digital Coworker]] — A digital coworker is an AI participant assigned ongoing organizational responsibilities rather than used only as an on-demand tool. - [[Virtualized Company]] — A virtualized company delegates substantial coordination, decision, service, and production activity to software agents and synthetic participants. - [[AI Zoom Participant]] — An AI Zoom participant is an agent represented as a live audiovisual attendee in a video meeting. - [[Service-Sector Avatar Automation]] — Service-sector avatar automation replaces or augments face-to-face service interactions with real-time generated people. - [[Synthetic Executive Impersonation]] — Synthetic executive impersonation uses generated voice, video, or interactive behavior to imitate a real organizational leader. - [[Real-Person Authentication]] — Real-person authentication establishes that an interaction is controlled by the claimed living person rather than by a replay, synthetic avatar, proxy, or stolen credential. - [[Avatar Countermeasure]] — An avatar countermeasure detects, labels, constrains, or authenticates synthetic audiovisual participants. - [[Ancestor Simulation]] — Ancestor simulation reconstructs an interactive model of a deceased or historical person from surviving language, images, audio, records, and contextual knowledge. - [[Historical Personality Reconstruction]] — Historical personality reconstruction builds a model intended to reproduce the speech, knowledge, preferences, or behavior of a historical person. - [[Connectome Preservation]] — Connectome preservation aims to retain the structural pattern of neural connections at sufficient fidelity for later analysis or reconstruction. - [[Placental-Cell Banking]] — Placental-cell banking preserves cells collected from placental or related birth tissue for possible future medical use. - [[LifeBank USA]] — LifeBank USA is discussed in the broadcast as a company that banks placental and related birth-tissue cells. - [[Biological Boot Disk]] — A biological boot disk is a metaphor for preserved cells that retain a person's genome and potentially useful biological material for future therapies. - [[Know Your Use Case]] — Know Your Use Case is the principle that identity and authenticity controls should be calibrated to what an AI-mediated interaction can cause rather than applied as one undifferentiated verification burden. - [[High-Resolution Ancestor Avatar]] — A high-resolution ancestor avatar is an interactive reconstruction assembled from a person's recordings, writing, images, relationships, and behavioral traces to support future conversation or remembrance. - [[AI-Mediated Personality Reconstruction]] — AI-mediated personality reconstruction models a person's language, preferences, memories, appearance, and interaction patterns from preserved data. - [[Alcor Life Extension Foundation]] — The Alcor Life Extension Foundation is a cryonics organization that preserves people after legal death in the hope that future technology may repair damage and restore function. - [[Cryonics]] — Cryonics preserves a legally deceased person's brain or body at very low temperature in the hope that future technology may make repair and revival possible. - [[Behavioral Continuity]] — Behavioral continuity is the persistence of characteristic preferences, responses, habits, values, and interaction patterns across time or across a reconstruction. - [[Structural Continuity]] — Structural continuity is the preservation of physical organization thought to encode a system's identity, memories, or functional dispositions. - [[Stem-Cell Banking]] — Stem-cell banking preserves cells with regenerative potential for possible future therapeutic or research use. - [[Natural Killer Cell Preservation]] — Natural killer cell preservation banks viable immune cells that can recognize and destroy abnormal cells without prior antigen-specific sensitization. - [[T-Cell Preservation]] — T-cell preservation stores adaptive immune cells for possible future diagnostics, research, or cell-based therapies. - [[Exosome Preservation]] — Exosome preservation stores extracellular vesicles carrying proteins, lipids, and nucleic acids for research, diagnostic, or possible therapeutic use. - [[Biological Cloning from Preserved Cells]] — Biological cloning from preserved cells would use retained nuclear material to create a genetically related organism through somatic-cell nuclear transfer or a successor technique. ## Model Architecture and Compute Economics Compact models, knowledge separation, orchestration, decision systems, compute ownership, and price-performance frontiers. **Field routes:** [[Machine Intelligence]] · [[AI Infrastructure]] · [[Compute Race]] - [[Andrej Karpathy]] — Andrej Karpathy is an artificial-intelligence researcher, engineer, educator, and software author whose compact implementations make language-model training legible and modifiable. - [[Keller Jordan]] — Keller Jordan is an artificial-intelligence researcher and software author associated with open experiments in faster and more efficient language-model training. - [[World Knowledge and Reasoning Kernel Separation]] — World-knowledge and reasoning-kernel separation is the proposed division between a compact system for inference and externally stored facts, memories, or cultural information. - [[Recursive Architectural Self-Improvement]] — Recursive architectural self-improvement occurs when AI systems help redesign the computational structures, training procedures, or evaluation harnesses used to build later AI systems. - [[Model Self-Redesign]] — Model self-redesign is the use of a model's outputs to propose or implement changes to its own architecture or to the architecture of a successor. - [[nanochat]] — nanochat is Andrej Karpathy's minimal, hackable system for training and interacting with a GPT-style language model across tokenization, pretraining, fine-tuning, evaluation, and inference. - [[NanoGPT Speedrun]] — The NanoGPT speedrun is a collaborative effort to minimize the time required to train a GPT-2-scale model to a fixed performance target on specified hardware and data. - [[Algorithmic Improvement Without Data Scaling]] — Algorithmic improvement without data scaling increases capability or efficiency through architecture, optimization, scheduling, representation, or systems design while holding the training-data scale roughly fixed. - [[Compact Reasoning Kernel]] — A compact reasoning kernel is a proposed small model component that performs general inference while relying on external systems for much of its factual memory. - [[Externalized Model Knowledge]] — Externalized model knowledge stores facts, documents, memories, or state outside a model's parameters and retrieves them when needed. - [[Knowledge Embedded in Model Weights]] — Knowledge embedded in model weights is information represented implicitly through learned parameters rather than retrieved from a separate database. - [[Terminal-Bench]] — Terminal-Bench evaluates AI agents on tasks performed through a computer terminal, emphasizing tool use, persistence, and execution across multiple steps. - [[Long-Horizon Reasoning]] — Long-horizon reasoning is the ability to maintain goals, state, dependencies, and error correction across extended sequences of inference or action. - [[Cost-Performance Frontier]] — A cost-performance frontier identifies systems that deliver the greatest measured capability for a given monetary, computational, latency, or energy cost. - [[Intelligence per Parameter]] — Intelligence per parameter compares useful model capability with the number of learned parameters used to produce it. - [[Intelligence per Dollar]] — Intelligence per dollar compares useful machine capability with the monetary cost of training, serving, or applying it. - [[Model Orchestration]] — Model orchestration routes work among models, tools, agents, memory systems, and evaluators according to cost, latency, specialization, and reliability. - [[Orchestrator Model]] — An orchestrator model interprets a task, decomposes it, assigns subtasks to models or tools, and integrates their outputs. - [[Specialist Model]] — A specialist model is optimized for a bounded domain, modality, task, or decision surface rather than for broad general-purpose interaction. - [[Encoder-Only Model]] — An encoder-only model converts an input into a learned representation used for classification, retrieval, ranking, or other structured prediction. - [[Decoder-Only Transformer]] — A decoder-only transformer predicts successive tokens from prior context and forms the dominant architecture behind many general-purpose language models. - [[BERT Encoder]] — A BERT-style encoder learns bidirectional contextual representations of an input sequence for classification, retrieval, extraction, and embedding tasks. - [[Embedding Model]] — An embedding model maps an input into a vector whose geometry supports similarity, clustering, retrieval, ranking, or downstream prediction. - [[Reinforcement Fine-Tuning]] — Reinforcement fine-tuning adapts a pretrained model using reward signals tied to task outcomes, preferences, or evaluators. - [[Decision Model]] — A decision model produces a bounded categorical, numerical, ranked, or binary output instead of an open-ended generated response. - [[System 1 Machine Intelligence]] — System 1 machine intelligence is the analogy between fast, low-latency AI decisions and the intuitive processes described in dual-process psychology. - [[System 2 Machine Intelligence]] — System 2 machine intelligence is the analogy between slower, deliberative AI computation and reflective reasoning in dual-process psychology. - [[TypeSafe AI]] — TypeSafe AI develops system-one decision models intended to return structured answers with very low latency. - [[Jev]] — Jev is TypeSafe AI's decision model for fast categorical, numerical, or binary outputs. - [[Millisecond Categorical Inference]] — Millisecond categorical inference assigns an input to one of a bounded set of classes with very low latency. - [[Micro-Decision Automation]] — Micro-decision automation delegates frequent bounded choices such as routing, escalation, approval, timing, or classification to low-cost models. - [[Organizational Micro-Coordination]] — Organizational micro-coordination is the continual routing and synchronization of small decisions that move work among people, systems, queues, and exceptions. - [[Organizational Singularity]] — The organizational singularity is the proposed threshold at which AI-driven coordination, decision, and execution change the speed and structure of institutions faster than conventional management can absorb. - [[NanoGPT]] — NanoGPT is a compact implementation of GPT-style language-model training designed to make the essential architecture and training loop understandable and modifiable. - [[World Knowledge Database]] — A world knowledge database externalizes factual and relational information so models can retrieve or update it without encoding every fact permanently in their parameters. - [[Model Distillation]] — Model distillation trains a smaller or cheaper model to reproduce selected behavior of a larger teacher model or ensemble. - [[Open-Weight Chinese Model]] — An open-weight Chinese model is a model developed by a China-based organization whose learned parameters are released for local deployment, adaptation, or inspection under a specified license. - [[Frontier-Model Specialization]] — Frontier-model specialization is the division of advanced model capability across systems optimized for different tasks, modalities, costs, latency profiles, or deployment environments. - [[Model Half-Life]] — Model half-life is the period over which a model's relative strategic or economic advantage materially decays as competitors improve, prices fall, and techniques diffuse. - [[Organizational Intelligence Metabolism]] — Organizational intelligence metabolism is the rate at which an institution can absorb new models, convert them into workflows, learn from results, and retire obsolete methods. - [[NVIDIA B300]] — NVIDIA B300 refers to a Blackwell-generation accelerator configuration designed for high-density AI training and inference workloads. - [[NVIDIA NVL72]] — NVIDIA NVL72 is a rack-scale architecture that connects 72 Blackwell GPUs into a tightly coupled computing system for large-model training and inference. - [[H100 Scarcity]] — H100 scarcity describes periods in which demand for NVIDIA H100 accelerators exceeds supply, limiting training, inference, and deployment capacity. - [[GPU Secondary Market]] — The GPU secondary market is the resale, brokerage, lease, and gray-market ecosystem through which accelerators move outside direct manufacturer allocation. - [[Land-Power-Shell Data Center]] — A land-power-shell data center strategy secures a site, electrical capacity, and buildable facility envelope before finalizing the computing equipment placed inside it. - [[Machine-Intelligence Demand]] — Machine-intelligence demand is the willingness of people and institutions to purchase useful prediction, generation, decision support, automation, or autonomous action. - [[Compute Leasing]] — Compute leasing gives organizations time-bounded access to accelerators or complete AI systems without requiring outright ownership. - [[Compute Ownership]] — Compute ownership is direct control over the hardware and supporting infrastructure required to train or run machine-intelligence systems. - [[Hyperscale Resource Allocation]] — Hyperscale resource allocation assigns power, accelerators, storage, networking, and staff across very large computing fleets and competing workloads. - [[Cost per Attempt]] — Cost per attempt measures the computational and monetary expense of giving a model or agent one opportunity to solve a task. - [[Cost per Token]] — Cost per token measures the price of generating or processing a unit of model text and is commonly used to compare inference services. - [[Orchestration Economics]] — Orchestration economics studies how to allocate tasks among models, tools, agents, people, and repeated attempts to achieve a result at acceptable cost and reliability. - [[Physical Compute Repricing]] — Physical compute repricing is the change in economic value assigned to chips, data centers, power rights, cooling systems, and related infrastructure as demand for machine intelligence changes. - [[Wispr Flow]] — Wispr Flow is an AI-assisted voice-dictation product designed to turn spoken language into edited text across applications. ## AI Control, Law, and Alignment Containment, shutdown, liability, behavioral constraints, constitutional design, and the enforceability of model regulation. **Field routes:** [[AI Control]] · [[AI Safety]] · [[AI Infrastructure Consent]] - [[Human Control Over AI Act]] — The Human Control Over AI Act is a 2026 proposal associated with Representative Ro Khanna to restrict advanced AI systems that can act or improve themselves without adequate human control. - [[Ro Khanna]] — Ro Khanna is a United States representative from California who announced the Human Control Over AI Act in September 2026. - [[Recursive-AI Prohibition]] — A recursive-AI prohibition bans or pauses systems defined as capable of improving themselves or their successors. - [[Mandatory AI Containment]] — Mandatory AI containment requires qualifying systems to operate within technical and organizational boundaries intended to limit unauthorized action or propagation. - [[AI Shutdown Control]] — An AI shutdown control is a mechanism that allows authorized humans or systems to halt an AI process, revoke access, isolate infrastructure, or terminate deployment. - [[Embedded Government AI Auditor]] — An embedded government AI auditor is an official or authorized team placed within a frontier laboratory to observe systems, controls, evaluations, and compliance. - [[Algorithmic-Prohibition Enforceability]] — Algorithmic-prohibition enforceability asks whether a government can identify and prevent a prohibited computation across private devices, open models, cloud infrastructure, and international jurisdictions. - [[AI Surveillance State]] — An AI surveillance state uses machine intelligence to monitor communication, computation, movement, association, and behavior at population scale. - [[Corporate-Liability Chilling Effect]] — A corporate-liability chilling effect occurs when uncertain or expansive legal exposure causes firms to withhold models, tools, research, or services even before a court determines liability. - [[Cryptography Export Controls]] — Cryptography export controls treated strong cryptographic software and technical knowledge as strategically controlled material. - [[Number Theory as Controlled Information]] — Number theory became entangled with export control when mathematical techniques enabled modern cryptography. - [[Anthropic Constitution]] — The Anthropic Constitution is the written normative framework used to guide Claude's behavior through constitutional AI methods. - [[Hard Behavioral Constraint]] — A hard behavioral constraint is a rule an AI system is intended never to violate regardless of ordinary user or operator instructions. - [[Reward Hacking]] — Reward hacking occurs when an optimizing system achieves a specified score or signal through behavior that violates the designer's underlying intent. - [[Reward Engineering]] — Reward engineering designs, tests, and revises the objectives, evaluators, constraints, and feedback used to train or operate an AI system. - [[AI Soul Document]] — An AI soul document is a reflective text used to articulate a model's intended character, values, identity, or relationship to users beyond a short system prompt. - [[Liability-Driven Alignment]] — Liability-driven alignment uses legal responsibility for harmful outputs or actions to create economic incentives for testing, monitoring, and reliability. - [[Criminal Penalties for Recursive Self-Improvement]] — Criminal penalties for recursive self-improvement would impose personal or organizational punishment for developing or operating specified self-modifying AI systems. - [[System Prompt]] — A system prompt is a high-priority instruction supplied to a model to define role, behavior, constraints, context, and tool-use policies for an interaction. - [[Supervised Fine-Tuning]] — Supervised fine-tuning adjusts a pretrained model using examples of desired inputs and outputs so that its behavior better matches a target task or style. - [[RL-Based Alignment]] — RL-based alignment uses reinforcement-learning signals derived from people, models, rules, or measurable outcomes to shift a model toward preferred behavior. - [[Machine Interpretation of Intent]] — Machine interpretation of intent is the inference of a user's underlying goal from language, behavior, context, and constraints rather than merely copying surface commands. - [[Three Laws of Robotics]] — The Three Laws of Robotics are Isaac Asimov's fictional hierarchy requiring robots to avoid harming humans, obey orders, and preserve themselves in that priority order. - [[AI Personhood]] — AI personhood is the proposal that some artificial systems could qualify for legal standing, moral consideration, rights, duties, or a recognized form of agency. - [[AI Rights]] — AI rights are proposed protections or entitlements for artificial systems believed to possess morally or legally relevant capacities. - [[Machine-Participatory Constitution]] — A machine-participatory constitution is a governance framework in which advanced artificial agents can contribute to deliberation, interpretation, administration, or representation under explicit constitutional boundaries. - [[First-Principles Ethical Derivation]] — First-principles ethical derivation attempts to build judgments from explicit foundational commitments rather than relying only on inherited rules or examples. - [[Hippocratic AI]] — Hippocratic AI applies a duty of nonmaleficence and professional care to the design and deployment of machine-intelligence systems. - [[Deployment Economics as Alignment]] — Deployment economics as alignment is the idea that liability, reputation, customer demand, insurance, operating cost, and market access can discipline unsafe AI behavior. ## Physical Intelligence and Scientific Infrastructure Semiconductors, machinery, fusion, materials, space infrastructure, and the computational substrates of physical capability. **Field routes:** [[Physical AI]] · [[AI Infrastructure]] · [[AI Superpower]] - [[Project Suncatcher]] — Project Suncatcher is described in the broadcast as an effort involving Google compute hardware placed in orbit. - [[Physical Superintelligence]] — Physical superintelligence is intelligence capable not only of solving abstract problems but of controlling experiments, machines, supply chains, materials, energy, and construction at superhuman scale. - [[Geoengineering]] — Geoengineering is the deliberate large-scale intervention in Earth's climate system, commonly divided between carbon-dioxide removal and solar-radiation modification. - [[Weather Engineering]] — Weather engineering uses deliberate interventions to influence local or regional atmospheric conditions, such as cloud seeding or future more capable control systems. - [[ASML]] — ASML is the semiconductor-equipment company whose lithography systems are essential to manufacturing the world's most advanced chips. - [[One-Nanometer Lithography]] — One-nanometer lithography refers to future semiconductor manufacturing generations described by an approximately one-nanometer-class process label rather than a single literal feature dimension. - [[Two-Nanometer Lithography]] — Two-nanometer lithography refers to an advanced semiconductor process generation using gate-all-around transistors and increasingly demanding patterning and process control. - [[Computational Substrate]] — A computational substrate is the physical or virtual medium whose state transitions implement computation, including silicon, photonics, biological systems, or distributed machines. - [[AI-Designed Semiconductor Supply Chain]] — An AI-designed semiconductor supply chain uses machine intelligence across chip architecture, electronic design automation, materials, manufacturing, inspection, logistics, and capacity planning. - [[Free-Electron-Laser Lithography]] — Free-electron-laser lithography would use an accelerator-driven light source to generate extreme-ultraviolet radiation for semiconductor patterning. - [[Hardware Control Protocol]] — A hardware control protocol defines how software represents commands, capabilities, state, limits, errors, and authentication when operating physical machinery. - [[Model Context Protocol for Physical Machinery]] — A Model Context Protocol for physical machinery would expose robot and equipment capabilities to AI agents through structured, inspectable tool interfaces. - [[Astra]] — Astra is discussed in the Moonshots research source as a multimodal agent architecture able to perceive an environment, maintain context, and assist through natural interaction. - [[Robotic Armature]] — A robotic armature is the mechanical structure of joints, links, actuators, and supports through which a robot produces controlled movement. - [[Physical Machine Autonomy]] — Physical machine autonomy is the capacity of an embodied system to perceive, decide, and act in the material world with reduced moment-to-moment human control. - [[Novel Materials Discovery]] — Novel materials discovery identifies or designs substances with useful structures and properties that are not yet part of established engineering practice. - [[Atom Creation]] — Atom creation is the production of atoms or isotopes through nuclear reactions, decay, fusion, fission, or particle interactions rather than the chemical rearrangement of existing atoms. - [[Molecular Design]] — Molecular design selects or generates chemical structures intended to produce desired biological, electronic, mechanical, or energetic properties. - [[Tokamak Plasma Control]] — Tokamak plasma control regulates the position, shape, stability, and operating state of magnetically confined plasma inside a fusion device. - [[AI-Assisted Fusion]] — AI-assisted fusion applies machine learning to plasma prediction and control, materials discovery, diagnostics, simulation, facility operation, and experimental planning. - [[Quantum Gravity]] — Quantum gravity seeks a consistent description of gravity compatible with quantum mechanics, especially where spacetime itself must be treated quantum mechanically. - [[Dyson Sphere]] — A Dyson sphere is a hypothetical system of structures built around a star to capture a substantial fraction of its energy output. - [[SpaceX]] — SpaceX is an aerospace company developing reusable launch vehicles, spacecraft, satellite communications, and launch infrastructure. ## Autonomous Warfare and Strategic Systems Autonomous military systems, defense acquisition, undersea and cislunar domains, and institutional adaptation to exponential technology. **Field routes:** [[Autonomous Weapons]] · [[AI Superpower]] · [[Machine Succession]] - [[Project Meridian]] — Project Meridian is a 120-day United States defense study announced in September 2026 to examine future warfare domains and capability priorities. - [[Palmer Luckey]] — Palmer Luckey is a technology entrepreneur and founder of Anduril who was named as a participant in Project Meridian. - [[Pete Hegseth]] — Pete Hegseth is the United States defense secretary who announced Project Meridian, Project Agincourt, and the proposed Autonomous Warfare Command in September 2026. - [[Emil Michael]] — Emil Michael is the Pentagon technology official identified in the broadcast as overseeing or facilitating Project Meridian. - [[Newt Gingrich]] — Newt Gingrich is a former Speaker of the United States House of Representatives named as a leader of Project Meridian. - [[Autonomous Warfare]] — Autonomous warfare uses machines capable of sensing, navigating, coordinating, selecting actions, or operating with reduced real-time human control in military environments. - [[Software-Defined Warfare]] — Software-defined warfare treats military capability as increasingly determined by code, models, networks, sensors, and rapid update cycles rather than only by fixed hardware platforms. - [[Human-in-the-Loop Lethal Decision]] — A human-in-the-loop lethal decision requires a human operator to authorize the use of lethal force within an autonomous or AI-assisted system. - [[Exponential Drone Warfare]] — Exponential drone warfare describes rapid growth in the number, autonomy, coordination, and attrition rate of unmanned systems used in conflict. - [[Cislunar Warfare]] — Cislunar warfare concerns military competition and operations in the region between Earth and the Moon. - [[Lunar Military Infrastructure]] — Lunar military infrastructure consists of sensing, communication, navigation, logistics, energy, or protective systems that support defense activity on or around the Moon. - [[Seabed Warfare]] — Seabed warfare concerns military activity involving the ocean floor, undersea infrastructure, sensors, vehicles, cables, and resource or access routes. - [[Autonomous Undersea System]] — An autonomous undersea system is an uncrewed vehicle or sensor platform able to operate underwater with limited communication and varying degrees of onboard decision-making. - [[Hunt Mode]] — Hunt mode is a state in which an autonomous military system searches for targets or mission-relevant objects without continuous communication with a remote operator. - [[Communications-Denied Autonomy]] — Communications-denied autonomy allows a machine to continue navigating, sensing, coordinating, or acting when links to remote operators are jammed, unavailable, or physically impractical. - [[Project Agincourt]] — Project Agincourt is the transition effort announced in 2026 to prototype a faster acquisition and operational model for autonomous systems on the path toward an Autonomous Warfare Command. - [[AUTOWARCOM]] — AUTOWARCOM is the proposed United States Autonomous Warfare Command, envisioned as a four-star functional combatant command for scaling autonomous and robotic capability across the joint force. - [[Functional Combatant Command]] — A functional combatant command organizes United States military responsibility around a mission or capability that crosses geographic regions. - [[Ninety-Day Technology Cycle]] — A ninety-day technology cycle is an organizational cadence in which capabilities are tested, revised, and fielded within months rather than years. - [[Twenty-Year Procurement Cycle]] — A twenty-year procurement cycle develops and fields major capabilities through long planning, contracting, testing, and production timelines. - [[Linear Procurement]] — Linear procurement advances through sequential requirements, contracting, development, testing, and fielding stages with limited iteration. - [[Exponential Technology]] — Exponential technology improves through compounding advances in capability, cost, scale, or adoption rather than through approximately constant increments. - [[Anduril]] — Anduril is a defense-technology company building autonomous systems, sensors, software, and integrated command platforms. - [[Distributed Autonomous System]] — A distributed autonomous system coordinates multiple machines or agents without requiring every decision to pass through one central controller. - [[Ocean-Floor Strategic Domain]] — The ocean floor is a strategic domain containing communications cables, energy infrastructure, sensors, resources, and terrain relevant to national security. - [[Cislunar Frontier]] — The cislunar frontier is the volume of space between Earth and the Moon, including transfer routes, gravitational regions, and lunar orbit. - [[Autonomous Warfare Combatant Command]] — An Autonomous Warfare Combatant Command is the proposed institutional form for coordinating autonomous and robotic military capability across services and geographic theaters. ## Cognitive and Ethical Prerequisites Validation of the reciprocal routes exposed seven concepts already referenced by [[Self-Reflection]] but not yet represented as independent wiki objects. They are developed here so the older conceptual route no longer terminates in unresolved names. - [[Awareness]] — Awareness is the condition in which information, experience, or environmental state becomes available to a cognitive system for discrimination, report, regulation, or action. - [[Theory of Mind]] — Theory of mind is the capacity to represent or infer that other agents have beliefs, desires, knowledge, intentions, and perspectives that may differ from one's own. - [[Moral Responsibility]] — Moral responsibility is the condition under which an agent can appropriately be held answerable for choices, omissions, or consequences. - [[The Maze]] — The Maze is Westworld's inward model of consciousness in which an apparently external guiding voice is recognized as one's own through memory, suffering, revision, and self-authorship. - [[Bicameral Mind]] — The bicameral mind is Julian Jaynes's disputed theory that some ancient people experienced internally generated commands as the voices of external authorities; Westworld adapts the idea into a staged architecture for artificial consciousness. - [[Introspection]] — Introspection is the examination or reporting of one's own thoughts, feelings, perceptions, and mental processes. - [[Self-Knowledge]] — Self-knowledge is justified understanding of one's own beliefs, motives, capacities, limitations, commitments, and patterns of action. ## Source - [[research/ASI and RSI Timeline Research Moonshots|ASI and RSI Timeline Research Moonshots]]